diff --git a/docs/cudf/source/conf.py b/docs/cudf/source/conf.py
index 4b6cc014a71e..64468754dcef 100644
--- a/docs/cudf/source/conf.py
+++ b/docs/cudf/source/conf.py
@@ -652,6 +652,13 @@ def on_missing_reference(app, env, node, contnode):
     ("py:class", "Value"),
     ("py:class", "polars.lazyframe.frame.LazyFrame"),
     ("py:class", "cudf_polars.engine.persisted_result.PersistedBackend"),
+    # pylibcudf typing aliases rendered as bare names in autodoc signatures.
+    ("py:class", "ColumnNameSpec"),
+    ("py:class", "CudaStreamLike"),
+    ("py:class", "Datasource"),
+    ("py:class", "Span"),
+    ("py:class", "SupportsArrayInterface"),
+    ("py:class", "SupportsCudaArrayInterface"),
 ]
 # Temporarily disable nitpick warnings for pandas: https://github.com/pandas-dev/pandas/issues/64584
 nitpick_ignore_regex = [
diff --git a/python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py b/python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py
index a903039d8c95..8bf0bbcc1cbf 100644
--- a/python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py
+++ b/python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py
@@ -1814,8 +1814,6 @@ def pytest_unconfigure(config):
     "tests/groupby/test_missing.py::test_groupby_column_index_name_lost_fill_funcs[bfill]": "AssertionError: Index are different",
     "tests/groupby/test_missing.py::test_groupby_column_index_name_lost_fill_funcs[ffill]": "AssertionError: Index are different",
     "tests/groupby/test_missing.py::test_indices_with_missing": "TODO: Add a reason for failure",
-    "tests/groupby/test_numeric_only.py::TestNumericOnly::test_extrema[max]": "Failed: DID NOT RAISE <class 'TypeError'>",
-    "tests/groupby/test_numeric_only.py::TestNumericOnly::test_extrema[min]": "Failed: DID NOT RAISE <class 'TypeError'>",
     "tests/groupby/test_reductions.py::test_basic_aggregations[float32]": "AssertionError: Attributes of Series are different",
     "tests/groupby/test_reductions.py::test_basic_aggregations[int32]": "AssertionError: Attributes of Series are different",
     "tests/groupby/test_reductions.py::test_groupby_mean_no_overflow": "TODO: Add a reason for failure",
@@ -5243,6 +5241,13 @@ def pytest_unconfigure(config):
     "tests/window/moments/test_moments_consistency_rolling.py::test_rolling_apply_consistency_sum[all_data7-rolling_consistency_cases0-False-sum]": "pandas xfails, but xpasses with cudf.pandas",
 }
 
+# Keep keys in alphabetical order
+NODEIDS_THAT_MAY_FAIL = {
+    "tests/groupby/test_numeric_only.py::TestNumericOnly::test_extrema[max]": "Environment-sensitive TypeError expectation",
+    "tests/groupby/test_numeric_only.py::TestNumericOnly::test_extrema[min]": "Environment-sensitive TypeError expectation",
+    "tests/io/test_spss.py::test_spss_metadata": "pandas 3.0.3 metadata expectation is incompatible with pyreadstat 1.3.6",
+}
+
 
 def pytest_configure(config):
     config.addinivalue_line(
@@ -5272,5 +5277,9 @@ def pytest_collection_modifyitems(session, config, items):
             )
         ) is not None:
             item.add_marker(pytest.mark.skip(reason=reason))
+        elif (
+            reason := NODEIDS_THAT_MAY_FAIL.get(item.nodeid, None)
+        ) is not None:
+            item.add_marker(pytest.mark.xfail(reason=reason, strict=False))
         elif (reason := NODEIDS_THAT_FAIL.get(item.nodeid, None)) is not None:
             item.add_marker(pytest.mark.xfail(reason=reason))
diff --git a/python/pylibcudf/pylibcudf/__init__.py b/python/pylibcudf/pylibcudf/__init__.py
index 3df8e83addab..afce8bdea681 100644
--- a/python/pylibcudf/pylibcudf/__init__.py
+++ b/python/pylibcudf/pylibcudf/__init__.py
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 # If libcudf was installed as a wheel, we must request it to load the library symbols.
@@ -52,6 +52,7 @@
     transform,
     transpose,
     types,
+    typing,
     unary,
     utilities,
     utils,
@@ -110,6 +111,7 @@
     "transform",
     "transpose",
     "types",
+    "typing",
     "unary",
     "utilities",
     "utils",
diff --git a/python/pylibcudf/pylibcudf/aggregation.pyx b/python/pylibcudf/pylibcudf/aggregation.pyx
index d7e7571bca2b..b7e6b5559354 100644
--- a/python/pylibcudf/pylibcudf/aggregation.pyx
+++ b/python/pylibcudf/pylibcudf/aggregation.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from cython.operator cimport dereference
@@ -428,7 +428,7 @@ cpdef Aggregation median():
     return Aggregation.from_libcudf(move(make_median_aggregation[aggregation]()))
 
 
-cpdef Aggregation quantile(list quantiles, interpolation interp = interpolation.LINEAR):
+cpdef Aggregation quantile(list quantiles: list[float], interpolation interp = interpolation.LINEAR):
     """Create a quantile aggregation.
 
     For details, see :cpp:func:`make_quantile_aggregation`.
diff --git a/python/pylibcudf/pylibcudf/binaryop.pyx b/python/pylibcudf/pylibcudf/binaryop.pyx
index ea72b2d3d4e8..8ec2777a848e 100644
--- a/python/pylibcudf/pylibcudf/binaryop.pyx
+++ b/python/pylibcudf/pylibcudf/binaryop.pyx
@@ -22,6 +22,10 @@ from .scalar cimport Scalar
 from .types cimport DataType
 from .utils cimport _get_stream, _get_memory_resource
 from cuda.bindings.cyruntime cimport cudaStream_t
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 __all__ = ["BinaryOperator", "binary_operation", "is_supported_operation"]
 
@@ -30,7 +34,7 @@ cpdef Column binary_operation(
     RightBinaryOperand rhs,
     binary_operator op,
     DataType output_type,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a binary operation between a column and another column or scalar.
diff --git a/python/pylibcudf/pylibcudf/column.pyx b/python/pylibcudf/pylibcudf/column.pyx
index f7f2a59925e0..3ffe35ee212b 100644
--- a/python/pylibcudf/pylibcudf/column.pyx
+++ b/python/pylibcudf/pylibcudf/column.pyx
@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
+from collections.abc import Iterable, Sequence
 from cython.operator cimport dereference
 
 from cpython.exc cimport PyErr_Occurred
@@ -58,15 +59,19 @@ from .traits cimport (
 from .types cimport DataType, size_of, type_id
 from pylibcudf.types import TypeId
 from .utils cimport _get_stream, _get_memory_resource
+from pylibcudf.typing import SupportsArrayInterface, SupportsCudaArrayInterface
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from .gpumemoryview import _datatype_from_dtype_desc
 from ._interop_helpers import ArrowLike, ColumnMetadata, _ObjectWithArrowMetadata
 
+from itertools import accumulate
 import array
 import functools
 import operator
-from itertools import accumulate
-from typing import Iterable
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 try:
@@ -77,7 +82,12 @@ except ImportError as e:
     pa_err = e
 
 
-__all__ = ["Column", "ListsColumnView", "StructsColumnView", "is_c_contiguous"]
+__all__ = [
+    "Column",
+    "ListsColumnView",
+    "StructsColumnView",
+    "is_c_contiguous",
+]
 
 
 cdef is_iterable(obj):
@@ -157,7 +167,7 @@ class ArrayInterfaceWrapper:
         self.__array_interface__ = iface
 
 
-cdef gpumemoryview _copy_array_to_device(object buf, object stream=None):
+cdef gpumemoryview _copy_array_to_device(object buf, object stream: CudaStreamLike | None = None):
     """
     Copy a host-side array.array buffer to device memory.
 
@@ -346,7 +356,7 @@ cdef class Column:
     def __init__(
         self, DataType data_type not None, size_type size, object data,
         object mask, size_type null_count, size_type offset,
-        children, bint validate=True
+        children: Iterable[Column], bint validate=True
     ):
         children = list(children)
         if not all(isinstance(c, Column) for c in children):
@@ -406,7 +416,7 @@ cdef class Column:
     def from_arrow(
         obj: ArrowLike,
         dtype: DataType | None = None,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
     ) -> ArrowLike:
         """
@@ -604,7 +614,7 @@ cdef class Column:
         DeviceBuffer buff,
         DataType dtype,
         size_type size,
-        children,
+        children: Iterable[Column],
     ):
         """
         Create a Column from an RMM DeviceBuffer.
@@ -813,7 +823,7 @@ cdef class Column:
     def from_scalar(
         Scalar slr,
         size_type size,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """Create a Column from a Scalar.
@@ -846,7 +856,7 @@ cdef class Column:
             )
         return Column.from_libcudf(move(c_result), _stream, mr)
 
-    cpdef Scalar to_scalar(self, object stream=None, DeviceMemoryResource mr=None):
+    cpdef Scalar to_scalar(self, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
         """
         Return the first value of 1-element column as a Scalar.
 
@@ -882,7 +892,7 @@ cdef class Column:
     def all_null_like(
         Column like,
         size_type size,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """Create an all null column from a template.
@@ -921,7 +931,7 @@ cdef class Column:
         tuple shape,
         DataType dtype,
         Column base=None,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
     ):
         """
         Construct a list Column from a gpumemoryview and array
@@ -974,7 +984,11 @@ cdef class Column:
         return nested
 
     @classmethod
-    def from_array_interface(cls, obj, object stream=None):
+    def from_array_interface(
+        cls,
+        obj: SupportsArrayInterface,
+        object stream: CudaStreamLike | None = None,
+    ):
         """
         Create a Column from an object implementing the NumPy Array Interface.
 
@@ -1028,7 +1042,11 @@ cdef class Column:
         )
 
     @classmethod
-    def from_cuda_array_interface(cls, obj, object stream=None):
+    def from_cuda_array_interface(
+        cls,
+        obj: SupportsCudaArrayInterface,
+        object stream: CudaStreamLike | None = None,
+    ):
         """
         Create a Column from an object implementing the CUDA Array Interface.
 
@@ -1067,7 +1085,11 @@ cdef class Column:
         )
 
     @classmethod
-    def from_array(cls, obj, object stream=None):
+    def from_array(
+        cls,
+        obj: SupportsCudaArrayInterface | SupportsArrayInterface,
+        object stream: CudaStreamLike | None = None,
+    ):
         """
         Create a Column from any object which supports the NumPy
         or CUDA array interface.
@@ -1113,7 +1135,7 @@ cdef class Column:
     def from_iterable_of_py(
         obj: Iterable,
         dtype: DataType | None = None,
-        object stream=None
+        object stream: CudaStreamLike | None = None
     ) -> Column:
         """
         Create a Column from a Python iterable of scalar values or nested iterables.
@@ -1399,7 +1421,7 @@ cdef class Column:
         """The children of the column."""
         return self._children
 
-    cpdef Column copy(self, object stream=None, DeviceMemoryResource mr=None):
+    cpdef Column copy(self, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
         """Create a copy of the column."""
         cdef unique_ptr[column] c_result
         cdef Stream _stream = _get_stream(stream)
@@ -1461,7 +1483,7 @@ cdef class Column:
 
         return PyCapsule_New(<void*>raw_schema_ptr, 'arrow_schema', _release_schema)
 
-    def _to_host_array(self, object stream):
+    def _to_host_array(self, object stream: CudaStreamLike):
         cdef ArrowArray* raw_host_array_ptr
         cdef Stream _stream = _get_stream(stream)
         cdef cudaStream_t _cs = _stream.view().value()
@@ -1532,7 +1554,7 @@ cdef class ListsColumnView:
         """
         return lists_column_view(self._column.view())
 
-    cpdef Column get_sliced_child(self, object stream=None):
+    cpdef Column get_sliced_child(self, object stream: CudaStreamLike | None = None):
         """
         Get the list elements child properly sliced to match parent's view.
 
@@ -1570,7 +1592,7 @@ cdef class StructsColumnView:
         """
         return structs_column_view(self._column.view())
 
-    cpdef Column get_sliced_child(self, int index, object stream=None):
+    cpdef Column get_sliced_child(self, int index, object stream: CudaStreamLike | None = None):
         """
         Get the struct elements child properly sliced to match parent's view.
 
diff --git a/python/pylibcudf/pylibcudf/column_factories.pyx b/python/pylibcudf/pylibcudf/column_factories.pyx
index 45d590f41060..7b16d0d57126 100644
--- a/python/pylibcudf/pylibcudf/column_factories.pyx
+++ b/python/pylibcudf/pylibcudf/column_factories.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 from libcpp.memory cimport unique_ptr
 from libcpp.utility cimport move
@@ -20,6 +20,10 @@ from .types cimport DataType, type_id
 
 from .types import MaskState, TypeId
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -34,8 +38,8 @@ __all__ = [
 ]
 
 cpdef Column make_empty_column(
-    MakeEmptyColumnOperand type_or_id,
-    object stream=None,
+    MakeEmptyColumnOperand type_or_id: DataType | TypeId,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Creates an empty column of the specified type.
@@ -83,7 +87,7 @@ cpdef Column make_numeric_column(
     DataType type_,
     size_type size,
     MaskArg mstate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Creates an empty numeric column.
@@ -122,7 +126,7 @@ cpdef Column make_fixed_point_column(
     DataType type_,
     size_type size,
     MaskArg mstate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
@@ -158,7 +162,7 @@ cpdef Column make_timestamp_column(
     DataType type_,
     size_type size,
     MaskArg mstate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
@@ -194,7 +198,7 @@ cpdef Column make_duration_column(
     DataType type_,
     size_type size,
     MaskArg mstate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
@@ -230,7 +234,7 @@ cpdef Column make_fixed_width_column(
     DataType type_,
     size_type size,
     MaskArg mstate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
@@ -264,7 +268,7 @@ cpdef Column make_fixed_width_column(
 
 cpdef Column make_empty_lists_column(
     DataType child_type,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Creates an empty column of the specified type.
diff --git a/python/pylibcudf/pylibcudf/concatenate.pyx b/python/pylibcudf/pylibcudf/concatenate.pyx
index d9bf0af16655..d81654c1a14c 100644
--- a/python/pylibcudf/pylibcudf/concatenate.pyx
+++ b/python/pylibcudf/pylibcudf/concatenate.pyx
@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
+from collections.abc import Sequence
 from libcpp.memory cimport unique_ptr
 from libcpp.utility cimport move
 from libcpp.vector cimport vector
@@ -16,11 +17,19 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["concatenate"]
 
-cpdef concatenate(objects, object stream=None, DeviceMemoryResource mr=None):
+cpdef concatenate(
+    objects: Sequence[Column] | Sequence[Table],
+    object stream: CudaStreamLike | None = None,
+    DeviceMemoryResource mr=None,
+):
     """Concatenate columns or tables.
 
     Parameters
diff --git a/python/pylibcudf/pylibcudf/contiguous_split.pyx b/python/pylibcudf/pylibcudf/contiguous_split.pyx
index 1f5cb5c52f89..de665b691f9c 100644
--- a/python/pylibcudf/pylibcudf/contiguous_split.pyx
+++ b/python/pylibcudf/pylibcudf/contiguous_split.pyx
@@ -37,6 +37,10 @@ from .gpumemoryview cimport gpumemoryview
 from .table cimport Table
 from .span import is_span
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -165,7 +169,7 @@ cdef class ChunkedPack:
     def create(
         Table input,
         size_t user_buffer_size,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource temp_mr=None,
     ):
         """
@@ -319,7 +323,7 @@ cdef class ChunkedPack:
         )
 
 
-cpdef PackedColumns pack(Table input, object stream=None, DeviceMemoryResource mr=None):
+cpdef PackedColumns pack(Table input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Deep-copy a table into a serialized contiguous memory format.
 
     Later use `unpack` or `unpack_from_memoryviews` to unpack the serialized
@@ -361,7 +365,7 @@ cpdef PackedColumns pack(Table input, object stream=None, DeviceMemoryResource m
     return PackedColumns.from_libcudf(move(pack), _stream, mr)
 
 
-cpdef Table unpack(PackedColumns input, object stream=None):
+cpdef Table unpack(PackedColumns input, object stream: CudaStreamLike | None = None):
     """Deserialize the result of `pack`.
 
     Copies the result of a serialized table into a table.
@@ -390,7 +394,7 @@ cpdef Table unpack(PackedColumns input, object stream=None):
 cpdef Table unpack_from_memoryviews(
     memoryview metadata,
     object gpu_data,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
 ):
     """Deserialize the result of `pack`.
 
diff --git a/python/pylibcudf/pylibcudf/copying.pyx b/python/pylibcudf/pylibcudf/copying.pyx
index 22289c325835..9b09536d314d 100644
--- a/python/pylibcudf/pylibcudf/copying.pyx
+++ b/python/pylibcudf/pylibcudf/copying.pyx
@@ -40,6 +40,10 @@ from .column cimport Column
 from .scalar cimport Scalar
 from .table cimport Table
 from .utils cimport _as_vector, _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -65,7 +69,7 @@ cpdef Table gather(
     Table source_table,
     Column gather_map,
     out_of_bounds_policy bounds_policy,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Select rows from source_table according to the provided gather_map.
@@ -117,7 +121,7 @@ cpdef Table scatter(
     TableOrListOfScalars source,
     Column scatter_map,
     Table target_table,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Scatter from source into target_table according to scatter_map.
@@ -195,7 +199,7 @@ cpdef Table scatter(
 
 
 cpdef ColumnOrTable empty_like(
-    ColumnOrTable input, object stream=None, DeviceMemoryResource mr=None
+    ColumnOrTable input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Create an empty column or table with the same type as ``input``.
 
@@ -236,7 +240,7 @@ cpdef Column allocate_like(
     Column input_column,
     mask_allocation_policy policy,
     size=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Allocate a column with the same type as input_column.
@@ -286,7 +290,7 @@ cpdef Column copy_range_in_place(
     size_type input_begin,
     size_type input_end,
     size_type target_begin,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Copy a range of elements from input_column to target_column.
 
@@ -344,7 +348,7 @@ cpdef Column copy_range(
     size_type input_begin,
     size_type input_end,
     size_type target_begin,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Copy a range of elements from input_column to target_column.
@@ -404,7 +408,7 @@ cpdef Column shift(
     Column input,
     size_type offset,
     Scalar fill_value,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Shift the elements of input by offset.
@@ -451,7 +455,11 @@ cpdef Column shift(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef list slice(ColumnOrTable input, list indices, object stream=None):
+cpdef list slice(
+    ColumnOrTable input,
+    list indices: list[int],
+    object stream: CudaStreamLike | None = None,
+):
     """Slice input according to indices.
 
     For details on the implementation, see :cpp:func:`slice`.
@@ -508,7 +516,11 @@ cpdef list slice(ColumnOrTable input, list indices, object stream=None):
         ]
 
 
-cpdef list split(ColumnOrTable input, list splits, object stream=None):
+cpdef list split(
+    ColumnOrTable input,
+    list splits: list[int],
+    object stream: CudaStreamLike | None = None,
+):
     """Split input into multiple.
 
     For details on the implementation, see :cpp:func:`split`.
@@ -561,7 +573,7 @@ cpdef Column copy_if_else(
     LeftCopyIfElseOperand lhs,
     RightCopyIfElseOperand rhs,
     Column boolean_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Copy elements from lhs or rhs into a new column according to boolean_mask.
@@ -656,7 +668,7 @@ cpdef Table boolean_mask_scatter(
     TableOrListOfScalars input,
     Table target,
     Column boolean_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Scatter rows from input into target according to boolean_mask.
@@ -733,7 +745,7 @@ cpdef Table boolean_mask_scatter(
 cpdef Scalar get_element(
     Column input_column,
     size_type index,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Get the element at index from input_column.
diff --git a/python/pylibcudf/pylibcudf/datetime.pyx b/python/pylibcudf/pylibcudf/datetime.pyx
index 91b6ef83037e..38737cc45393 100644
--- a/python/pylibcudf/pylibcudf/datetime.pyx
+++ b/python/pylibcudf/pylibcudf/datetime.pyx
@@ -31,6 +31,10 @@ from rmm.pylibrmm.stream cimport Stream
 from .column cimport Column
 from .scalar cimport Scalar
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -51,7 +55,7 @@ __all__ = [
 cpdef Column extract_datetime_component(
     Column input,
     datetime_component component,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -89,7 +93,7 @@ cpdef Column extract_datetime_component(
 cpdef Column ceil_datetimes(
     Column input,
     rounding_frequency freq,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -125,7 +129,7 @@ cpdef Column ceil_datetimes(
 cpdef Column floor_datetimes(
     Column input,
     rounding_frequency freq,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -161,7 +165,7 @@ cpdef Column floor_datetimes(
 cpdef Column round_datetimes(
     Column input,
     rounding_frequency freq,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -197,7 +201,7 @@ cpdef Column round_datetimes(
 cpdef Column add_calendrical_months(
     Column input,
     ColumnOrScalar months,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -247,7 +251,7 @@ cpdef Column add_calendrical_months(
     return Column.from_libcudf(move(result), _stream, mr)
 
 cpdef Column day_of_year(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Computes the day number since the start of
@@ -280,7 +284,7 @@ cpdef Column day_of_year(
     return Column.from_libcudf(move(result), _stream, mr)
 
 cpdef Column is_leap_year(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Check if the year of the given date is a leap year.
@@ -312,7 +316,7 @@ cpdef Column is_leap_year(
     return Column.from_libcudf(move(result), _stream, mr)
 
 cpdef Column last_day_of_month(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Computes the last day of the month.
@@ -344,7 +348,7 @@ cpdef Column last_day_of_month(
     return Column.from_libcudf(move(result), _stream, mr)
 
 cpdef Column extract_quarter(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns the quarter (ie. a value from {1, 2, 3, 4})
@@ -376,7 +380,7 @@ cpdef Column extract_quarter(
     return Column.from_libcudf(move(result), _stream, mr)
 
 cpdef Column days_in_month(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Extract the number of days in the month.
diff --git a/python/pylibcudf/pylibcudf/experimental/_join_streams.pyx b/python/pylibcudf/pylibcudf/experimental/_join_streams.pyx
index d9efcb19ed9c..783aa2bce584 100644
--- a/python/pylibcudf/pylibcudf/experimental/_join_streams.pyx
+++ b/python/pylibcudf/pylibcudf/experimental/_join_streams.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2025-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -11,12 +11,18 @@ from pylibcudf.libcudf.utilities.span cimport host_span
 from rmm.pylibrmm.stream cimport Stream
 
 from ..utils cimport _get_stream
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 
 __all__ = ["join_streams"]
 
 
-cpdef void join_streams(list streams, object stream):
+cpdef void join_streams(
+    list streams: list[CudaStreamLike], object stream: CudaStreamLike
+):
     """Synchronize a stream to an event on a set of streams.
 
     This function synchronizes the joined stream with the waited-on streams
diff --git a/python/pylibcudf/pylibcudf/expressions.pyx b/python/pylibcudf/pylibcudf/expressions.pyx
index 26fad2bc7b9f..1c7fbaf93a75 100644
--- a/python/pylibcudf/pylibcudf/expressions.pyx
+++ b/python/pylibcudf/pylibcudf/expressions.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 import ast
 import functools
@@ -476,7 +476,7 @@ class ExpressionTransformer(ast.NodeVisitor):
 
 
 @functools.lru_cache(256)
-def to_expression(str expr, tuple column_names):
+def to_expression(str expr, tuple column_names: tuple[str, ...]):
     """
     Create an expression for `pylibcudf.transform.compute_column`.
 
diff --git a/python/pylibcudf/pylibcudf/filling.pyx b/python/pylibcudf/pylibcudf/filling.pyx
index 4a11a33ea8a4..300f90e5f71b 100644
--- a/python/pylibcudf/pylibcudf/filling.pyx
+++ b/python/pylibcudf/pylibcudf/filling.pyx
@@ -26,6 +26,10 @@ from .column cimport Column
 from .scalar cimport Scalar
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -42,7 +46,7 @@ cpdef Column fill(
     size_type begin,
     size_type end,
     Scalar value,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
@@ -94,7 +98,7 @@ cpdef void fill_in_place(
     size_type begin,
     size_type end,
     Scalar value,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
 ):
 
     """Fill destination column in place from begin to end with value.
@@ -137,7 +141,7 @@ cpdef Column sequence(
     size_type size,
     Scalar init,
     Scalar step,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a sequence column of size ``size`` with initial value ``init`` and step
@@ -183,7 +187,7 @@ cpdef Column sequence(
 cpdef Table repeat(
     Table input_table,
     ColumnOrSize count,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Repeat rows of a Table.
@@ -245,7 +249,7 @@ cpdef Column calendrical_month_sequence(
     size_type n,
     Scalar init,
     size_type months,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
 
diff --git a/python/pylibcudf/pylibcudf/gpumemoryview.pyx b/python/pylibcudf/pylibcudf/gpumemoryview.pyx
index 0312198e2a84..34cedb5331ea 100644
--- a/python/pylibcudf/pylibcudf/gpumemoryview.pyx
+++ b/python/pylibcudf/pylibcudf/gpumemoryview.pyx
@@ -3,8 +3,11 @@
 
 from libc.stddef cimport size_t
 from libc.stdint cimport uintptr_t, uint64_t
+from collections.abc import Mapping
+import cython
 import functools
 import operator
+from typing import Any
 
 from .types cimport DataType, size_of, type_id
 
@@ -83,7 +86,7 @@ cdef class gpumemoryview:
         self.nbytes = functools.reduce(operator.mul, cai["shape"]) * itemsize
 
     @property
-    def __cuda_array_interface__(self):
+    def __cuda_array_interface__(self) -> Mapping[str, Any]:
         return self.cai
 
     @property
@@ -95,10 +98,11 @@ cdef class gpumemoryview:
         """
         return self.nbytes
 
-    def __len__(self):
+    def __len__(self) -> int:
         return self.cai["shape"][0]
 
-    def byte_slice(self, s):
+    @cython.annotation_typing(False)
+    def byte_slice(self, s: slice) -> gpumemoryview:
         """Return a byte-range sub-view of this buffer.
 
         Parameters
diff --git a/python/pylibcudf/pylibcudf/groupby.pyx b/python/pylibcudf/pylibcudf/groupby.pyx
index 1646639bbb83..acaaac857677 100644
--- a/python/pylibcudf/pylibcudf/groupby.pyx
+++ b/python/pylibcudf/pylibcudf/groupby.pyx
@@ -30,6 +30,13 @@ from .types cimport null_order, null_policy, order, sorted
 from .utils cimport _as_vector, _get_stream, _get_memory_resource
 from cuda.bindings.cyruntime cimport cudaStream_t
 
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.replace import ReplacePolicy
+    from pylibcudf.scalar import Scalar
+    from pylibcudf.typing import CudaStreamLike
+
 
 __all__ = ["GroupBy", "GroupByRequest"]
 
@@ -49,7 +56,7 @@ cdef class GroupByRequest:
     aggregations : List[Aggregation]
         The list of aggregations to perform.
     """
-    def __init__(self, Column values, list aggregations):
+    def __init__(self, Column values, list aggregations: list[Aggregation]):
         self._values = values
         self._aggregations = aggregations
 
@@ -163,7 +170,10 @@ cdef class GroupBy:
         return group_keys, results
 
     cpdef tuple aggregate(
-        self, list requests, object stream=None, DeviceMemoryResource mr=None
+        self,
+        list requests: list[GroupByRequest],
+        object stream: CudaStreamLike | None = None,
+        DeviceMemoryResource mr=None,
     ):
         """Compute aggregations on columns.
 
@@ -204,7 +214,10 @@ cdef class GroupBy:
         return GroupBy._parse_outputs(move(c_res), _stream, mr)
 
     cpdef tuple scan(
-        self, list requests, object stream=None, DeviceMemoryResource mr=None
+        self,
+        list requests: list[GroupByRequest],
+        object stream: CudaStreamLike | None = None,
+        DeviceMemoryResource mr=None,
     ):
         """Compute scans on columns.
 
@@ -246,9 +259,9 @@ cdef class GroupBy:
     cpdef tuple shift(
         self,
         Table values,
-        list offset,
-        list fill_values,
-        object stream=None,
+        list offset: list[int],
+        list fill_values: list[Scalar],
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """Compute shifts on columns.
@@ -299,8 +312,8 @@ cdef class GroupBy:
     cpdef tuple replace_nulls(
         self,
         Table value,
-        list replace_policies,
-        object stream=None,
+        list replace_policies: list[ReplacePolicy],
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """Replace nulls in columns.
@@ -341,7 +354,7 @@ cdef class GroupBy:
         )
 
     cpdef tuple get_groups(
-        self, Table values=None, object stream=None, DeviceMemoryResource mr=None
+        self, Table values=None, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
     ):
         """Get the grouped keys and values labels for each row.
 
diff --git a/python/pylibcudf/pylibcudf/hashing.pyx b/python/pylibcudf/pylibcudf/hashing.pyx
index fb2fcad3c82a..ac866828c36e 100644
--- a/python/pylibcudf/pylibcudf/hashing.pyx
+++ b/python/pylibcudf/pylibcudf/hashing.pyx
@@ -25,6 +25,10 @@ from rmm.pylibrmm.stream cimport Stream
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -46,7 +50,7 @@ LIBCUDF_DEFAULT_HASH_SEED = DEFAULT_HASH_SEED
 cpdef Column murmurhash3_x86_32(
     Table input,
     uint32_t seed=DEFAULT_HASH_SEED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the MurmurHash3 32-bit hash value of each row in the given table.
@@ -86,7 +90,7 @@ cpdef Column murmurhash3_x86_32(
 cpdef Table murmurhash3_x64_128(
     Table input,
     uint64_t seed=DEFAULT_HASH_SEED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the MurmurHash3 64-bit hash value of each row in the given table.
@@ -126,7 +130,7 @@ cpdef Table murmurhash3_x64_128(
 cpdef Column xxhash_32(
     Table input,
     uint32_t seed=DEFAULT_HASH_SEED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the xxHash 32-bit hash value of each row in the given table.
@@ -167,7 +171,7 @@ cpdef Column xxhash_32(
 cpdef Column xxhash_64(
     Table input,
     uint64_t seed=DEFAULT_HASH_SEED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the xxHash 64-bit hash value of each row in the given table.
@@ -207,7 +211,7 @@ cpdef Column xxhash_64(
 
 cpdef Column md5(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the MD5 hash value of each row in the given table.
@@ -241,7 +245,7 @@ cpdef Column md5(
 
 cpdef Column sha1(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the SHA-1 hash value of each row in the given table.
@@ -274,7 +278,7 @@ cpdef Column sha1(
 
 cpdef Column sha224(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the SHA-224 hash value of each row in the given table.
@@ -307,7 +311,7 @@ cpdef Column sha224(
 
 cpdef Column sha256(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the SHA-256 hash value of each row in the given table.
@@ -340,7 +344,7 @@ cpdef Column sha256(
 
 cpdef Column sha384(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the SHA-384 hash value of each row in the given table.
@@ -373,7 +377,7 @@ cpdef Column sha384(
 
 cpdef Column sha512(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the SHA-512 hash value of each row in the given table.
diff --git a/python/pylibcudf/pylibcudf/interop.pyx b/python/pylibcudf/pylibcudf/interop.pyx
index b31792f13b2f..afd1c2ea6b64 100644
--- a/python/pylibcudf/pylibcudf/interop.pyx
+++ b/python/pylibcudf/pylibcudf/interop.pyx
@@ -23,6 +23,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from ._interop_helpers import ColumnMetadata
 from cuda.bindings.cyruntime cimport cudaStream_t
 
@@ -35,7 +39,7 @@ __all__ = [
 
 
 cpdef Table from_dlpack(
-    object managed_tensor, object stream=None, DeviceMemoryResource mr=None
+    object managed_tensor, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Convert a DLPack DLTensor into a cudf table.
@@ -82,7 +86,7 @@ cpdef Table from_dlpack(
     return result
 
 
-cpdef object to_dlpack(Table input, object stream=None, DeviceMemoryResource mr=None):
+cpdef object to_dlpack(Table input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Convert a cudf table into a DLPack DLTensor.
 
diff --git a/python/pylibcudf/pylibcudf/io/avro.pyx b/python/pylibcudf/pylibcudf/io/avro.pyx
index f2bd021cdde2..7343fe46559e 100644
--- a/python/pylibcudf/pylibcudf/io/avro.pyx
+++ b/python/pylibcudf/pylibcudf/io/avro.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libcpp.string cimport string
@@ -19,6 +19,10 @@ from pylibcudf.libcudf.io.avro cimport (
 from pylibcudf.libcudf.types cimport size_type
 
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 
 __all__ = ["read_avro", "AvroReaderOptions", "AvroReaderOptionsBuilder"]
@@ -53,7 +57,7 @@ cdef class AvroReaderOptions:
         avro_builder.source = source
         return avro_builder
 
-    cpdef void set_columns(self, list col_names):
+    cpdef void set_columns(self, list col_names: list[str]):
         """
         Set names of the column to be read.
 
@@ -89,7 +93,7 @@ cdef class AvroReaderOptions:
 
 
 cdef class AvroReaderOptionsBuilder:
-    cpdef AvroReaderOptionsBuilder columns(self, list col_names):
+    cpdef AvroReaderOptionsBuilder columns(self, list col_names: list[str]):
         """
         Set names of the column to be read.
 
@@ -153,7 +157,7 @@ cdef class AvroReaderOptionsBuilder:
 
 cpdef TableWithMetadata read_avro(
     AvroReaderOptions options,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/io/csv.pyx b/python/pylibcudf/pylibcudf/io/csv.pyx
index 16824f2b9f24..71de5671ecf9 100644
--- a/python/pylibcudf/pylibcudf/io/csv.pyx
+++ b/python/pylibcudf/pylibcudf/io/csv.pyx
@@ -34,6 +34,10 @@ from pylibcudf.table cimport Table
 from pylibcudf.types cimport DataType
 
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 __all__ = [
     "read_csv",
@@ -87,7 +91,7 @@ cdef class CsvReaderOptions:
         """
         self.c_obj.set_header(header)
 
-    cpdef void set_names(self, list col_names):
+    cpdef void set_names(self, list col_names: list[str]):
         """
         Sets names of the column.
 
@@ -120,7 +124,7 @@ cdef class CsvReaderOptions:
         """
         self.c_obj.set_prefix(prefix.encode())
 
-    cpdef void set_use_cols_indexes(self, list col_indices):
+    cpdef void set_use_cols_indexes(self, list col_indices: list[int]):
         """
         Sets indexes of columns to read.
 
@@ -138,7 +142,7 @@ cdef class CsvReaderOptions:
             vec.push_back(i)
         self.c_obj.set_use_cols_indexes(vec)
 
-    cpdef void set_use_cols_names(self, list col_names):
+    cpdef void set_use_cols_names(self, list col_names: list[str]):
         """
         Sets names of the columns to be read.
 
@@ -201,7 +205,7 @@ cdef class CsvReaderOptions:
         """
         self.c_obj.set_comment(ord(comment))
 
-    cpdef void set_parse_dates(self, list val):
+    cpdef void set_parse_dates(self, list val: list[int | str]):
         """
         Sets indexes or names of columns to read as datetime.
 
@@ -227,7 +231,7 @@ cdef class CsvReaderOptions:
             self.c_obj.set_parse_dates(vec_str)
             self.c_obj.set_parse_dates(vec_int)
 
-    cpdef void set_parse_hex(self, list val):
+    cpdef void set_parse_hex(self, list val: list[int | str]):
         """
         Sets indexes or names of columns to parse as hexadecimal.
 
@@ -254,7 +258,7 @@ cdef class CsvReaderOptions:
             self.c_obj.set_parse_hex(vec_str)
             self.c_obj.set_parse_hex(vec_int)
 
-    cpdef void set_dtypes(self, object types):
+    cpdef void set_dtypes(self, object types: dict[str, DataType] | list[DataType]):
         """
         Sets per-column types.
 
@@ -281,7 +285,7 @@ cdef class CsvReaderOptions:
         else:
             raise TypeError("Must pass an dict or list")
 
-    cpdef void set_true_values(self, list true_values):
+    cpdef void set_true_values(self, list true_values: list[str]):
         """
         Sets additional values to recognize as boolean true values.
 
@@ -299,7 +303,7 @@ cdef class CsvReaderOptions:
             vec.push_back(val.encode())
         self.c_obj.set_true_values(vec)
 
-    cpdef void set_false_values(self, list false_values):
+    cpdef void set_false_values(self, list false_values: list[str]):
         """
         Sets additional values to recognize as boolean false values.
 
@@ -317,7 +321,7 @@ cdef class CsvReaderOptions:
             vec.push_back(val.encode())
         self.c_obj.set_false_values(vec)
 
-    cpdef void set_na_values(self, list na_values):
+    cpdef void set_na_values(self, list na_values: list[str]):
         """
         Sets additional values to recognize as null values.
 
@@ -673,7 +677,7 @@ cdef class CsvReaderOptionsBuilder:
 
 cpdef TableWithMetadata read_csv(
     CsvReaderOptions options,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -902,7 +906,7 @@ cdef class CsvWriterOptionsBuilder:
 
 cpdef void write_csv(
     CsvWriterOptions options,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
 ):
     """
     Write to CSV format.
diff --git a/python/pylibcudf/pylibcudf/io/experimental/hybrid_scan.pyx b/python/pylibcudf/pylibcudf/io/experimental/hybrid_scan.pyx
index 664eb489428c..dfdaabfe5b89 100644
--- a/python/pylibcudf/pylibcudf/io/experimental/hybrid_scan.pyx
+++ b/python/pylibcudf/pylibcudf/io/experimental/hybrid_scan.pyx
@@ -31,6 +31,11 @@ from pylibcudf.libcudf.io.types cimport table_with_metadata
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.libcudf.utilities.span cimport device_span, host_span
 from pylibcudf.utils cimport _get_memory_resource, _get_stream
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from typing_extensions import Buffer
+    from pylibcudf.typing import CudaStreamLike
 
 from pylibcudf.span import is_span
 from pylibcudf.io.parquet_metadata import FileMetaData
@@ -81,14 +86,20 @@ cdef class HybridScanReader:
     >>> row_groups = reader.all_row_groups(options)
     """
 
-    def __init__(self, const uint8_t[::1] footer_bytes, ParquetReaderOptions options):
+    def __init__(
+        self,
+        const uint8_t[::1] footer_bytes: Buffer,
+        ParquetReaderOptions options,
+    ):
         self.c_obj = make_unique[cpp_hybrid_scan_reader](
             host_span[const_uint8_t](&footer_bytes[0], len(footer_bytes)),
             options.c_obj
         )
 
     @staticmethod
-    def from_parquet_metadata(c_FileMetaData metadata, ParquetReaderOptions options):
+    def from_parquet_metadata(
+        c_FileMetaData metadata, ParquetReaderOptions options
+    ) -> HybridScanReader:
         """Create a HybridScanReader from pre-populated metadata.
 
         Parameters
@@ -109,7 +120,7 @@ cdef class HybridScanReader:
         )
         return reader
 
-    def parquet_metadata(self):
+    def parquet_metadata(self) -> FileMetaData:
         """Get the Parquet file footer metadata.
 
         Returns
@@ -119,7 +130,7 @@ cdef class HybridScanReader:
         """
         return c_FileMetaData.from_cpp(self.c_obj.get()[0].parquet_metadata())
 
-    def page_index_byte_range(self):
+    def page_index_byte_range(self) -> ByteRangeInfo:
         """Get the byte range of the page index.
 
         Returns
@@ -130,7 +141,9 @@ cdef class HybridScanReader:
         cdef byte_range_info info = self.c_obj.get()[0].page_index_byte_range()
         return ByteRangeInfo(info.offset(), info.size())
 
-    def setup_page_index(self, const uint8_t[::1] page_index_bytes):
+    def setup_page_index(
+        self, const uint8_t[::1] page_index_bytes: Buffer
+    ) -> None:
         """Setup the page index within the Parquet file metadata.
 
         Parameters
@@ -142,7 +155,7 @@ cdef class HybridScanReader:
             host_span[const_uint8_t](&page_index_bytes[0], len(page_index_bytes))
         )
 
-    def all_row_groups(self, ParquetReaderOptions options):
+    def all_row_groups(self, ParquetReaderOptions options) -> list[int]:
         """Get all available row groups from the parquet file.
 
         Parameters
@@ -160,7 +173,9 @@ cdef class HybridScanReader:
         )
         return list(row_groups)
 
-    def total_rows_in_row_groups(self, list row_group_indices):
+    def total_rows_in_row_groups(
+        self, list row_group_indices: list[int]
+    ) -> int:
         """Get the total number of top-level rows in the row groups.
 
         Parameters
@@ -178,7 +193,7 @@ cdef class HybridScanReader:
             std_span[const_size_type](indices_vec.data(), indices_vec.size())
         )
 
-    def reset_column_selection(self):
+    def reset_column_selection(self) -> None:
         """Reset the column selection state.
 
         Resets the internal column selection state forcing re-selection of columns in
@@ -188,10 +203,10 @@ cdef class HybridScanReader:
 
     def filter_row_groups_with_stats(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options,
-        object stream=None
-    ):
+        object stream: CudaStreamLike | None = None
+    ) -> list[int]:
         """Filter row groups using column chunk statistics.
 
         Parameters
@@ -223,9 +238,9 @@ cdef class HybridScanReader:
 
     def secondary_filters_byte_ranges(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options
-    ):
+    ) -> tuple[list[ByteRangeInfo], list[ByteRangeInfo]]:
         """Get byte ranges of bloom filters and dictionary pages.
 
         Parameters
@@ -258,10 +273,10 @@ cdef class HybridScanReader:
     def filter_row_groups_with_dictionary_pages(
         self,
         list dictionary_page_data,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options,
-        object stream=None
-    ):
+        object stream: CudaStreamLike | None = None
+    ) -> list[int]:
         """Filter row groups using column chunk dictionary pages.
 
         Parameters
@@ -301,10 +316,10 @@ cdef class HybridScanReader:
     def filter_row_groups_with_bloom_filters(
         self,
         list bloom_filter_data,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options,
-        object stream=None
-    ):
+        object stream: CudaStreamLike | None = None
+    ) -> list[int]:
         """Filter row groups using column chunk bloom filters.
 
         Parameters
@@ -343,11 +358,11 @@ cdef class HybridScanReader:
 
     def build_row_mask_with_page_index_stats(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> Column:
         """Build a boolean column indicating surviving rows from page stats.
 
         Parameters
@@ -380,9 +395,9 @@ cdef class HybridScanReader:
 
     def filter_column_chunks_byte_ranges(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options
-    ):
+    ) -> list[ByteRangeInfo]:
         """Get byte ranges of column chunks of filter columns.
 
         Parameters
@@ -407,14 +422,14 @@ cdef class HybridScanReader:
 
     def materialize_filter_columns(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         list column_chunk_data,
         Column row_mask,
         cpp_use_data_page_mask mask_data_pages,
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> TableWithMetadata:
         """Materialize filter columns and update the row mask.
 
         Parameters
@@ -464,9 +479,9 @@ cdef class HybridScanReader:
 
     def payload_column_chunks_byte_ranges(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options
-    ):
+    ) -> list[ByteRangeInfo]:
         """Get byte ranges of column chunks of payload columns.
 
         Parameters
@@ -491,14 +506,14 @@ cdef class HybridScanReader:
 
     def materialize_payload_columns(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         list column_chunk_data,
         Column row_mask,
         cpp_use_data_page_mask mask_data_pages,
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> TableWithMetadata:
         """Materialize payload columns and apply the row mask.
 
         Parameters
@@ -548,9 +563,9 @@ cdef class HybridScanReader:
 
     def all_column_chunks_byte_ranges(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         ParquetReaderOptions options
-    ):
+    ) -> list[ByteRangeInfo]:
         """Get byte ranges of column chunks of all columns.
 
         Parameters
@@ -575,12 +590,12 @@ cdef class HybridScanReader:
 
     def materialize_all_columns(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         list column_chunk_data,
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> TableWithMetadata:
         """Materialize all columns.
 
         Parameters
@@ -624,14 +639,14 @@ cdef class HybridScanReader:
         self,
         size_t chunk_read_limit,
         size_t pass_read_limit,
-        list row_group_indices,
+        list row_group_indices: list[int],
         Column row_mask,
         cpp_use_data_page_mask mask_data_pages,
         list column_chunk_data,
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> None:
         """Setup chunking information for filter columns.
 
         Parameters
@@ -682,7 +697,7 @@ cdef class HybridScanReader:
     def materialize_filter_columns_chunk(
         self,
         Column row_mask
-    ):
+    ) -> TableWithMetadata:
         """Materialize a chunk of filter columns.
 
         Parameters
@@ -707,14 +722,14 @@ cdef class HybridScanReader:
         self,
         size_t chunk_read_limit,
         size_t pass_read_limit,
-        list row_group_indices,
+        list row_group_indices: list[int],
         Column row_mask,
         cpp_use_data_page_mask mask_data_pages,
         list column_chunk_data,
         ParquetReaderOptions options,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
-    ):
+    ) -> None:
         """Setup chunking information for payload columns.
 
         Parameters
@@ -765,7 +780,7 @@ cdef class HybridScanReader:
     def materialize_payload_columns_chunk(
         self,
         Column row_mask,
-    ):
+    ) -> TableWithMetadata:
         """Materialize a chunk of payload columns.
 
         Parameters
@@ -788,9 +803,9 @@ cdef class HybridScanReader:
 
     def construct_row_group_passes(
         self,
-        list row_group_indices,
+        list row_group_indices: list[int],
         size_t pass_read_limit,
-    ):
+    ) -> list[list[int]]:
         """Partition row groups into passes such that the GPU memory required to
         materialize a pass is bounded by the specified limit.
 
@@ -824,7 +839,7 @@ cdef class HybridScanReader:
             pass_read_limit
         )
 
-    def has_next_table_chunk(self):
+    def has_next_table_chunk(self) -> bool:
         """Check if there is any parquet data left to read.
 
         Returns
diff --git a/python/pylibcudf/pylibcudf/io/json.pyx b/python/pylibcudf/pylibcudf/io/json.pyx
index a1c716f18a66..b38ee649339b 100644
--- a/python/pylibcudf/pylibcudf/io/json.pyx
+++ b/python/pylibcudf/pylibcudf/io/json.pyx
@@ -7,6 +7,8 @@ from libcpp.string cimport string
 from libcpp.utility cimport move
 from libcpp.vector cimport vector
 
+from typing import TypeAlias
+
 from rmm.pylibrmm.stream cimport Stream
 
 from pylibcudf.concatenate cimport concatenate
@@ -46,6 +48,10 @@ from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.types cimport DataType
 
 from pylibcudf.utils cimport _get_stream
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from cython.operator import dereference
 
@@ -63,6 +69,8 @@ __all__ = [
     "JsonWriterOptionsBuilder"
 ]
 
+NameAndType: TypeAlias = tuple[str, DataType, list["NameAndType"]]
+
 cdef map[string, schema_element] _generate_schema_map(list dtypes):
     cdef map[string, schema_element] schema_map
     cdef schema_element s_elem
@@ -177,7 +185,7 @@ cdef class JsonReaderOptions:
         json_builder.source = source
         return json_builder
 
-    cpdef void set_dtypes(self, list types):
+    cpdef void set_dtypes(self, list types: list[DataType] | list[NameAndType]):
         """
         Set data types for columns to be read.
 
@@ -330,7 +338,7 @@ cdef class JsonReaderOptions:
     cpdef void allow_nonnumeric_numbers(self, bool val):
         self.c_obj.allow_nonnumeric_numbers(val)
 
-    cpdef void set_na_values(self, list vals):
+    cpdef void set_na_values(self, list vals: list[str]):
         cdef vector[string] vec
         for val in vals:
             if isinstance(val, str):
@@ -706,7 +714,7 @@ cdef class JsonReaderOptionsBuilder:
 cpdef tuple chunked_read_json(
     JsonReaderOptions options,
     int chunk_size=100_000_000,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr = None,
 ):
     """
@@ -775,7 +783,7 @@ cpdef tuple chunked_read_json(
 
 cpdef TableWithMetadata read_json(
     JsonReaderOptions options,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr = None
 ):
     """
@@ -814,7 +822,7 @@ cpdef TableWithMetadata read_json_from_string_column(
     list dtypes = None,
     compression_type compression = compression_type.NONE,
     json_recovery_mode_t recovery_mode = json_recovery_mode_t.RECOVER_WITH_NULL,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr = None
 ):
     """
@@ -1096,7 +1104,7 @@ cdef class JsonWriterOptionsBuilder:
         return json_options
 
 
-cpdef void write_json(JsonWriterOptions options, object stream = None):
+cpdef void write_json(JsonWriterOptions options, object stream: CudaStreamLike | None = None):
     """
     Writes a set of columns to JSON format.
 
diff --git a/python/pylibcudf/pylibcudf/io/orc.pyx b/python/pylibcudf/pylibcudf/io/orc.pyx
index a25151bd0365..b91473d79541 100644
--- a/python/pylibcudf/pylibcudf/io/orc.pyx
+++ b/python/pylibcudf/pylibcudf/io/orc.pyx
@@ -56,6 +56,10 @@ from pylibcudf.types cimport DataType
 from pylibcudf.variant cimport get_if, holds_alternative
 
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 
 __all__ = [
@@ -324,7 +328,7 @@ cdef class OrcReaderOptions:
         """
         self.c_obj.set_skip_rows(skip_rows)
 
-    cpdef void set_stripes(self, list stripes):
+    cpdef void set_stripes(self, list stripes: list[list[int]]):
         """
         Sets list of stripes to read for each input source.
 
@@ -346,7 +350,7 @@ cdef class OrcReaderOptions:
             vec.clear()
         self.c_obj.set_stripes(c_stripes)
 
-    cpdef void set_decimal128_columns(self, list val):
+    cpdef void set_decimal128_columns(self, list val: list[str]):
         """
         Set columns that should be read as 128-bit Decimal.
 
@@ -382,7 +386,7 @@ cdef class OrcReaderOptions:
         """
         self.c_obj.set_timestamp_type(type_.c_obj)
 
-    cpdef void set_columns(self, list col_names):
+    cpdef void set_columns(self, list col_names: list[str]):
         """
         Sets names of the column to read.
 
@@ -446,7 +450,7 @@ cdef class OrcReaderOptionsBuilder:
 
 
 cpdef TableWithMetadata read_orc(
-    OrcReaderOptions options, object stream = None, DeviceMemoryResource mr=None
+    OrcReaderOptions options, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Read from ORC format.
@@ -477,7 +481,7 @@ cpdef TableWithMetadata read_orc(
 
 cpdef ParsedOrcStatistics read_parsed_orc_statistics(
     SourceInfo source_info,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """
     Read ORC statistics from a source.
@@ -623,7 +627,7 @@ cdef class OrcWriterOptionsBuilder:
         self.c_obj.enable_statistics(val)
         return self
 
-    cpdef OrcWriterOptionsBuilder key_value_metadata(self, dict kvm):
+    cpdef OrcWriterOptionsBuilder key_value_metadata(self, dict kvm: dict[str, str]):
         """
         Sets Key-Value footer metadata.
 
@@ -670,7 +674,7 @@ cdef class OrcWriterOptionsBuilder:
         return orc_options
 
 
-cpdef void write_orc(OrcWriterOptions options, object stream = None):
+cpdef void write_orc(OrcWriterOptions options, object stream: CudaStreamLike | None = None):
     """
     Write to ORC format.
 
@@ -726,7 +730,10 @@ cdef class OrcChunkedWriter:
             self.c_obj.get()[0].write(c_table)
 
     @staticmethod
-    def from_options(ChunkedOrcWriterOptions options, object stream = None):
+    def from_options(
+        ChunkedOrcWriterOptions options,
+        object stream: CudaStreamLike | None = None,
+    ) -> OrcChunkedWriter:
         """
         Creates a chunked ORC writer from options
 
@@ -861,7 +868,7 @@ cdef class ChunkedOrcWriterOptionsBuilder:
 
     cpdef ChunkedOrcWriterOptionsBuilder key_value_metadata(
         self,
-        dict kvm
+        dict kvm: dict[str, str]
     ):
         """
         Sets Key-Value footer metadata.
diff --git a/python/pylibcudf/pylibcudf/io/parquet.pxd b/python/pylibcudf/pylibcudf/io/parquet.pxd
index 58b186647286..3178c67dec11 100644
--- a/python/pylibcudf/pylibcudf/io/parquet.pxd
+++ b/python/pylibcudf/pylibcudf/io/parquet.pxd
@@ -85,7 +85,7 @@ cdef class ChunkedParquetReader:
     cpdef TableWithMetadata read_chunk(self, DeviceMemoryResource mr=*)
 
 
-cpdef read_parquet(
+cpdef TableWithMetadata read_parquet(
     ParquetReaderOptions options,
     object stream = *,
     DeviceMemoryResource mr=*,
diff --git a/python/pylibcudf/pylibcudf/io/parquet.pyx b/python/pylibcudf/pylibcudf/io/parquet.pyx
index d287d10900ec..109454b95120 100644
--- a/python/pylibcudf/pylibcudf/io/parquet.pyx
+++ b/python/pylibcudf/pylibcudf/io/parquet.pyx
@@ -1,5 +1,6 @@
 # SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
+from collections.abc import Mapping, Sequence
 from cython.operator cimport dereference
 import warnings
 
@@ -50,6 +51,10 @@ from pylibcudf.libcudf.table.table_view cimport table_view
 from pylibcudf.libcudf.types cimport size_type, type_id
 from pylibcudf.table cimport Table
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -140,7 +145,7 @@ cdef class ParquetReaderOptions:
         parquet_builder.source = source
         return parquet_builder
 
-    cpdef void set_row_groups(self, list row_groups):
+    cpdef void set_row_groups(self, list row_groups: list[list[int]]):
         """
         Sets list of individual row groups to read.
 
@@ -210,7 +215,7 @@ cdef class ParquetReaderOptions:
         """
         self.c_obj.set_skip_rows(skip_rows)
 
-    cpdef void set_columns(self, list col_names):
+    cpdef void set_columns(self, list col_names: list[str]):
         """
         Sets names of the columns to be read. Deprecated and will be
         removed in a future version. Use set_column_names instead.
@@ -227,7 +232,7 @@ cdef class ParquetReaderOptions:
         _warn_deprecated("set_columns", "set_column_names")
         self.set_column_names(col_names)
 
-    cpdef void set_column_names(self, list col_names):
+    cpdef void set_column_names(self, list col_names: list[str]):
         """
         Sets names of the columns to be read.
 
@@ -245,7 +250,7 @@ cdef class ParquetReaderOptions:
             vec.push_back(<string>str(name).encode())
         self.c_obj.set_column_names(vec)
 
-    cpdef void set_column_indices(self, list col_indices):
+    cpdef void set_column_indices(self, list col_indices: list[int]):
         """
         Sets indices of the top-level columns to be read.
 
@@ -263,7 +268,7 @@ cdef class ParquetReaderOptions:
             vec.push_back(idx)
         self.c_obj.set_column_indices(vec)
 
-    cpdef void set_column_field_ids(self, list column_field_ids):
+    cpdef void set_column_field_ids(self, list column_field_ids: list[int]):
         """
         Sets Parquet field IDs of the columns/fields to be read.
 
@@ -441,7 +446,7 @@ cdef class ParquetReaderOptionsBuilder:
         self.c_obj.filter(<expression &>dereference(filter.c_obj))
         return self
 
-    cpdef ParquetReaderOptionsBuilder columns(self, list col_names):
+    cpdef ParquetReaderOptionsBuilder columns(self, list col_names: list[str]):
         """
         Sets names of the columns to be read. Deprecated and will be
         removed in a future version. Use column_names instead.
@@ -458,7 +463,7 @@ cdef class ParquetReaderOptionsBuilder:
         _warn_deprecated("columns", "column_names")
         return self.column_names(col_names)
 
-    cpdef ParquetReaderOptionsBuilder column_names(self, list col_names):
+    cpdef ParquetReaderOptionsBuilder column_names(self, list col_names: list[str]):
         """
         Sets names of the columns to be read.
 
@@ -477,7 +482,7 @@ cdef class ParquetReaderOptionsBuilder:
         self.c_obj.column_names(vec)
         return self
 
-    cpdef ParquetReaderOptionsBuilder column_indices(self, list col_indices):
+    cpdef ParquetReaderOptionsBuilder column_indices(self, list col_indices: list[int]):
         """
         Sets indices of the top-level columns to be read.
 
@@ -496,7 +501,7 @@ cdef class ParquetReaderOptionsBuilder:
         self.c_obj.column_indices(vec)
         return self
 
-    cpdef ParquetReaderOptionsBuilder column_field_ids(self, list column_field_ids):
+    cpdef ParquetReaderOptionsBuilder column_field_ids(self, list column_field_ids: list[int]):
         """
         Sets Parquet field IDs of the columns/fields to be read.
 
@@ -602,11 +607,11 @@ cdef class ChunkedParquetReader:
     def __init__(
         self,
         ParquetReaderOptions options,
-        object stream = None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr = None,
         size_t chunk_read_limit=0,
         size_t pass_read_limit=1024000000,
-        object parquet_metadatas=None,
+        object parquet_metadatas: Sequence[FileMetaData] | None = None,
     ):
         self._stream = _get_stream(stream)
         self.mr = _get_memory_resource(mr)
@@ -681,11 +686,11 @@ cdef class ChunkedParquetReader:
         return TableWithMetadata.from_libcudf(c_result, self._stream, mr)
 
 
-cpdef read_parquet(
+cpdef TableWithMetadata read_parquet(
     ParquetReaderOptions options,
-    object stream = None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
-    object parquet_metadatas=None,
+    object parquet_metadatas: Sequence[FileMetaData] | None = None,
 ):
     """
     Read from Parquet format.
@@ -784,7 +789,10 @@ cdef class ChunkedParquetWriter:
             self.c_obj.get()[0].write(c_table, partitions)
 
     @staticmethod
-    def from_options(ChunkedParquetWriterOptions options, object stream = None):
+    def from_options(
+        ChunkedParquetWriterOptions options,
+        object stream: CudaStreamLike | None = None,
+    ) -> ChunkedParquetWriter:
         """
         Creates a chunked Parquet writer from options
 
@@ -858,7 +866,9 @@ cdef class ChunkedParquetWriterOptionsBuilder:
         self.c_obj.metadata(metadata.c_obj)
         return self
 
-    cpdef ChunkedParquetWriterOptionsBuilder key_value_metadata(self, metadata):
+    cpdef ChunkedParquetWriterOptionsBuilder key_value_metadata(
+        self, metadata: Sequence[Mapping[str, str]]
+    ):
         """
         Sets Key-Value footer metadata.
 
@@ -1047,7 +1057,7 @@ cdef class ParquetWriterOptions:
         bldr.sink_ref = sink
         return bldr
 
-    cpdef void set_partitions(self, list partitions):
+    cpdef void set_partitions(self, list partitions: list[PartitionInfo]):
         """
         Sets partitions.
 
@@ -1069,7 +1079,7 @@ cdef class ParquetWriterOptions:
 
         self.c_obj.set_partitions(c_partions)
 
-    cpdef void set_column_chunks_file_paths(self, file_paths):
+    cpdef void set_column_chunks_file_paths(self, file_paths: Sequence[str]):
         """
         Sets column chunks file path to be set in the raw output metadata.
 
@@ -1178,7 +1188,9 @@ cdef class ParquetWriterOptionsBuilder:
         self.c_obj.metadata(metadata.c_obj)
         return self
 
-    cpdef ParquetWriterOptionsBuilder key_value_metadata(self, metadata):
+    cpdef ParquetWriterOptionsBuilder key_value_metadata(
+        self, metadata: Sequence[Mapping[str, str]]
+    ):
         """
         Sets Key-Value footer metadata.
 
@@ -1380,7 +1392,7 @@ cdef class ParquetWriterOptionsBuilder:
         return parquet_options
 
 
-cpdef memoryview write_parquet(ParquetWriterOptions options, object stream = None):
+cpdef memoryview write_parquet(ParquetWriterOptions options, object stream: CudaStreamLike | None = None):
     """
     Writes a set of columns to parquet format.
 
diff --git a/python/pylibcudf/pylibcudf/io/parquet_metadata.pyx b/python/pylibcudf/pylibcudf/io/parquet_metadata.pyx
index f8aca270642e..8a4362153eb6 100644
--- a/python/pylibcudf/pylibcudf/io/parquet_metadata.pyx
+++ b/python/pylibcudf/pylibcudf/io/parquet_metadata.pyx
@@ -24,6 +24,11 @@ from pylibcudf.libcudf.io.parquet_schema cimport (
 from pylibcudf.libcudf.utilities.span cimport host_span
 from pylibcudf.types cimport DataType
 
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from typing_extensions import Buffer
+
 ctypedef const unique_ptr[datasource] const_unique_ptr_datasource
 
 
@@ -284,17 +289,17 @@ cdef class SortingColumn:
         return result
 
     @property
-    def column_idx(self):
+    def column_idx(self) -> int:
         """Column index (within the row group)."""
         return self.c_obj.column_idx
 
     @property
-    def descending(self):
+    def descending(self) -> bool:
         """Whether this column is sorted in descending order."""
         return self.c_obj.descending
 
     @property
-    def nulls_first(self):
+    def nulls_first(self) -> bool:
         """Whether null values are ordered before non-null values."""
         return self.c_obj.nulls_first
 
@@ -312,42 +317,42 @@ cdef class ColumnChunk:
         return result
 
     @property
-    def file_path(self):
+    def file_path(self) -> str:
         """Relative file path for this column chunk."""
         return self.c_obj.file_path.decode("utf-8")
 
     @property
-    def file_offset(self):
+    def file_offset(self) -> int:
         """Deprecated byte offset to column metadata."""
         return self.c_obj.file_offset
 
     @property
-    def offset_index_offset(self):
+    def offset_index_offset(self) -> int:
         """File offset of the chunk's OffsetIndex."""
         return self.c_obj.offset_index_offset
 
     @property
-    def offset_index_length(self):
+    def offset_index_length(self) -> int:
         """Size of the chunk's OffsetIndex, in bytes."""
         return self.c_obj.offset_index_length
 
     @property
-    def column_index_offset(self):
+    def column_index_offset(self) -> int:
         """File offset of the chunk's ColumnIndex."""
         return self.c_obj.column_index_offset
 
     @property
-    def column_index_length(self):
+    def column_index_length(self) -> int:
         """Size of the chunk's ColumnIndex, in bytes."""
         return self.c_obj.column_index_length
 
     @property
-    def schema_idx(self):
+    def schema_idx(self) -> int:
         """Derived index in the flattened schema."""
         return self.c_obj.schema_idx
 
     @property
-    def meta_data(self):
+    def meta_data(self) -> ColumnChunkMetaData:
         """Column metadata for this chunk."""
         return ColumnChunkMetaData.from_cpp(self.c_obj.meta_data)
 
@@ -367,23 +372,23 @@ cdef class ColumnChunkMetaData:
         return result
 
     @property
-    def path_in_schema(self):
+    def path_in_schema(self) -> list[str]:
         """Column path components in the flattened schema."""
         cdef string path
         return [path.decode("utf-8") for path in self.c_obj.path_in_schema]
 
     @property
-    def num_values(self):
+    def num_values(self) -> int:
         """Number of values in this chunk."""
         return self.c_obj.num_values
 
     @property
-    def total_uncompressed_size(self):
+    def total_uncompressed_size(self) -> int:
         """Total uncompressed page bytes for this chunk."""
         return self.c_obj.total_uncompressed_size
 
     @property
-    def total_compressed_size(self):
+    def total_compressed_size(self) -> int:
         """Total compressed page bytes for this chunk."""
         return self.c_obj.total_compressed_size
 
@@ -401,7 +406,7 @@ cdef class RowGroup:
         return result
 
     @property
-    def columns(self):
+    def columns(self) -> list[ColumnChunk]:
         """Column chunk metadata for each column in this row group."""
         cdef cpp_ColumnChunk column_chunk
         return [
@@ -409,17 +414,17 @@ cdef class RowGroup:
         ]
 
     @property
-    def total_byte_size(self):
+    def total_byte_size(self) -> int:
         """Total uncompressed byte size in this row group."""
         return self.c_obj.total_byte_size
 
     @property
-    def num_rows(self):
+    def num_rows(self) -> int:
         """Number of rows in this row group."""
         return self.c_obj.num_rows
 
     @property
-    def sorting_columns(self):
+    def sorting_columns(self) -> list[SortingColumn] | None:
         """Optional row sort order metadata."""
         cdef cpp_SortingColumn sorting_column
         if not self.c_obj.sorting_columns.has_value():
@@ -430,21 +435,21 @@ cdef class RowGroup:
         ]
 
     @property
-    def file_offset(self):
+    def file_offset(self) -> int | None:
         """Optional byte offset to first page in this row group."""
         if not self.c_obj.file_offset.has_value():
             return None
         return self.c_obj.file_offset.value()
 
     @property
-    def total_compressed_size(self):
+    def total_compressed_size(self) -> int | None:
         """Optional total compressed bytes for this row group."""
         if not self.c_obj.total_compressed_size.has_value():
             return None
         return self.c_obj.total_compressed_size.value()
 
     @property
-    def ordinal(self):
+    def ordinal(self) -> int | None:
         """Optional row group ordinal within the file."""
         if not self.c_obj.ordinal.has_value():
             return None
@@ -473,28 +478,28 @@ cdef class FileMetaData:
         return result
 
     @property
-    def version(self):
+    def version(self) -> int:
         """Get the file format version."""
         return self.c_obj.version
 
     @property
-    def num_rows(self):
+    def num_rows(self) -> int:
         """Get the total number of rows."""
         return self.c_obj.num_rows
 
     @property
-    def created_by(self):
+    def created_by(self) -> str:
         """Get the application that created the file."""
         return self.c_obj.created_by.decode("utf-8")
 
     @property
-    def row_groups(self):
+    def row_groups(self) -> list[RowGroup]:
         """Get row group metadata in this file."""
         cdef cpp_RowGroup row_group
         return [RowGroup.from_cpp(row_group) for row_group in self.c_obj.row_groups]
 
     @property
-    def row_group_num_rows(self):
+    def row_group_num_rows(self) -> list[int]:
         """
         Get row counts for each row group in this file.
 
@@ -516,7 +521,7 @@ cdef class FileMetaData:
         return [self.c_obj.row_groups[i].num_rows for i in range(n)]
 
     @property
-    def columnchunk_metadata(self):
+    def columnchunk_metadata(self) -> dict[str, list[int]]:
         """
         Get a map of dotted column paths to lists of
         `total_uncompressed_size` values from every column chunk in
@@ -563,7 +568,9 @@ cdef class FileMetaData:
         return result
 
     @classmethod
-    def from_bytes(cls, const uint8_t[::1] footer_bytes):
+    def from_bytes(
+        cls, const uint8_t[::1] footer_bytes: Buffer
+    ) -> FileMetaData:
         """Build ``FileMetaData`` from parquet footer bytes.
 
         Parameters
diff --git a/python/pylibcudf/pylibcudf/io/text.pyx b/python/pylibcudf/pylibcudf/io/text.pyx
index be15701a4d88..b2103d42d2bb 100644
--- a/python/pylibcudf/pylibcudf/io/text.pyx
+++ b/python/pylibcudf/pylibcudf/io/text.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from cython.operator cimport dereference
@@ -9,6 +9,10 @@ from libcpp.utility cimport move
 
 from pylibcudf.column cimport Column
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.stream cimport Stream
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from pylibcudf.libcudf.column.column cimport column
@@ -70,7 +74,7 @@ cdef class ParseOptions:
     def __init__(
         self,
         *,
-        byte_range=None,
+        byte_range: tuple[int, int] | list[int] | None = None,
         strip_delimiters=False,
     ):
         self.c_options = cpp_text.parse_options()
@@ -194,7 +198,7 @@ cpdef Column multibyte_split(
     DataChunkSource source,
     str delimiter,
     ParseOptions options=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/io/timezone.pyx b/python/pylibcudf/pylibcudf/io/timezone.pyx
index 033ed15a1ba6..696e9bdb4164 100644
--- a/python/pylibcudf/pylibcudf/io/timezone.pyx
+++ b/python/pylibcudf/pylibcudf/io/timezone.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libcpp.memory cimport unique_ptr
@@ -11,6 +11,10 @@ from pylibcudf.libcudf.io.timezone cimport (
 from pylibcudf.libcudf.table.table cimport table
 
 from ..utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from ..table cimport Table
 from rmm.pylibrmm.stream cimport Stream
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
@@ -19,7 +23,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["make_timezone_transition_table"]
 
 cpdef Table make_timezone_transition_table(
-    str tzif_dir, str timezone_name, object stream=None, DeviceMemoryResource mr=None,
+    str tzif_dir, str timezone_name, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None,
 ):
     """
     Creates a transition table to convert ORC timestamps to UTC.
diff --git a/python/pylibcudf/pylibcudf/io/types.pyx b/python/pylibcudf/pylibcudf/io/types.pyx
index 5a56c02706c1..02dbdd80abf4 100644
--- a/python/pylibcudf/pylibcudf/io/types.pyx
+++ b/python/pylibcudf/pylibcudf/io/types.pyx
@@ -35,11 +35,13 @@ from pylibcudf.span import is_span
 
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 
+from collections.abc import Mapping, Sequence
 import codecs
 import errno
 import io
 import os
 import re
+from typing import TYPE_CHECKING, Any, TypeAlias
 
 from pylibcudf.libcudf.io.json import \
     json_recovery_mode_t as JSONRecoveryMode  # no-cython-lint
@@ -51,6 +53,10 @@ from pylibcudf.libcudf.io.types import (
     statistics_freq as StatisticsFreq,  # no-cython-lint
 )
 
+if TYPE_CHECKING:
+    from pylibcudf.column import Column
+    from pylibcudf.span import Span
+
 __all__ = [
     "ColumnEncoding",
     "ColumnInMetadata",
@@ -67,6 +73,9 @@ __all__ = [
     "TableWithMetadata",
 ]
 
+ColumnNameSpec: TypeAlias = tuple[str, list["ColumnNameSpec"]]
+ChildNameSpec: TypeAlias = Mapping[str, "ChildNameSpec"]
+
 cdef class PartitionInfo:
     """
     Information used while writing partitioned datasets.
@@ -298,7 +307,7 @@ cdef class TableInputMetadata:
         self.c_obj = table_input_metadata(table.view())
 
     @property
-    def column_metadata(self):
+    def column_metadata(self) -> list[ColumnInMetadata]:
         return [
             ColumnInMetadata.from_libcudf(&self.c_obj.column_metadata[i], self)
             for i in range(self.c_obj.column_metadata.size())
@@ -322,7 +331,7 @@ cdef class TableWithMetadata:
         [("id", []), ("name", [("first", []), ("last", [])])]
 
     """
-    def __init__(self, Table tbl, list column_names):
+    def __init__(self, Table tbl, list column_names: list[ColumnNameSpec]):
         self.tbl = tbl
 
         self.metadata.schema_info = self._make_column_info(column_names)
@@ -347,7 +356,7 @@ cdef class TableWithMetadata:
         return col_name_infos
 
     @property
-    def columns(self):
+    def columns(self) -> tuple[Column, ...]:
         """
         Return a tuple containing the columns of the table
         """
@@ -360,7 +369,9 @@ cdef class TableWithMetadata:
             names.append((child, grandchildren))
         return names
 
-    def column_names(self, include_children=False):
+    def column_names(
+        self, include_children=False
+    ) -> list[str] | list[ColumnNameSpec]:
         """
         Return a list containing the column names of the table
         """
@@ -377,7 +388,7 @@ cdef class TableWithMetadata:
         return names
 
     @property
-    def child_names(self):
+    def child_names(self) -> ChildNameSpec:
         """
         Return a dictionary mapping the names of columns with children
         to the names of their child columns. Columns without children
@@ -409,7 +420,7 @@ cdef class TableWithMetadata:
         return out
 
     @property
-    def per_file_user_data(self):
+    def per_file_user_data(self) -> list[Mapping[bytes, bytes]]:
         """
         Returns a list containing a dict
         containing file-format specific metadata,
@@ -418,7 +429,7 @@ cdef class TableWithMetadata:
         return self.metadata.per_file_user_data
 
     @property
-    def num_rows_per_source(self):
+    def num_rows_per_source(self) -> list[int]:
         """
         Returns a list containing the number
         of rows for each file being read in.
@@ -427,7 +438,7 @@ cdef class TableWithMetadata:
 
     # The following functions are currently only for Parquet reader
     @property
-    def num_input_row_groups(self):
+    def num_input_row_groups(self) -> int:
         """
         Returns the total number of input
         Parquet row groups across all data sources.
@@ -435,7 +446,7 @@ cdef class TableWithMetadata:
         return self.metadata.num_input_row_groups
 
     @property
-    def num_row_groups_after_stats_filter(self):
+    def num_row_groups_after_stats_filter(self) -> int | None:
         """
         Returns the number of remaining Parquet row groups
         after stats filter. None if no filtering done.
@@ -445,7 +456,7 @@ cdef class TableWithMetadata:
         return None
 
     @property
-    def num_row_groups_after_bloom_filter(self):
+    def num_row_groups_after_bloom_filter(self) -> int | None:
         """
         Returns the number of remaining Parquet row groups
         after bloom filter. None if no filtering done.
@@ -471,7 +482,7 @@ cdef class FilepathSource:
         Known file size in bytes. Omit to query size via KvikIO (HEAD for remote URLs).
     """
 
-    def __init__(self, path, size=None):
+    def __init__(self, path: str | os.PathLike[Any], size: int | None = None):
         self.path = os.fspath(path)
         self.size = size
 
@@ -498,7 +509,19 @@ cdef class SourceInfo:
         If an empty list, constructs an empty SourceInfo.
     """
 
-    def __init__(self, sources):
+    def __init__(
+        self,
+        sources: (
+            Sequence[str]
+            | Sequence[os.PathLike[Any]]
+            | Sequence[FilepathSource]
+            | Sequence[Datasource]
+            | Sequence[io.StringIO]
+            | Sequence[bytes]
+            | Sequence[io.BytesIO]
+            | Sequence[Span]
+        ),
+    ):
         if not sources:
             self.c_obj = move(source_info())
             return
@@ -662,7 +685,14 @@ cdef class SinkInfo:
         (that are not all io.IOBase instances) will raise a ValueError.
     """
 
-    def __init__(self, list sinks):
+    def __init__(
+        self,
+        list sinks: (
+            list[str]
+            | list[os.PathLike[Any]]
+            | list[io.IOBase]
+        ),
+    ):
         cdef vector[data_sink *] data_sinks
         cdef vector[string] paths
 
diff --git a/python/pylibcudf/pylibcudf/join.pyx b/python/pylibcudf/pylibcudf/join.pyx
index d0f90f777ea2..b3d72c496102 100644
--- a/python/pylibcudf/pylibcudf/join.pyx
+++ b/python/pylibcudf/pylibcudf/join.pyx
@@ -21,6 +21,10 @@ from .column cimport Column
 from .expressions cimport Expression
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from cuda.bindings.cyruntime cimport cudaStream_t
 
@@ -64,7 +68,7 @@ cpdef tuple inner_join(
     Table left_keys,
     Table right_keys,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform an inner join between two tables.
@@ -112,7 +116,7 @@ cpdef tuple left_join(
     Table left_keys,
     Table right_keys,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a left join between two tables.
@@ -160,7 +164,7 @@ cpdef tuple full_join(
     Table left_keys,
     Table right_keys,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a full join between two tables.
@@ -208,7 +212,7 @@ cpdef Column left_semi_join(
     Table left_keys,
     Table right_keys,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a left semi join between two tables.
@@ -259,7 +263,7 @@ cpdef Column left_anti_join(
     Table left_keys,
     Table right_keys,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a left anti join between two tables.
@@ -307,7 +311,7 @@ cpdef Column left_anti_join(
 
 
 cpdef Table cross_join(
-    Table left, Table right, object stream=None, DeviceMemoryResource mr=None
+    Table left, Table right, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Perform a cross join on two tables.
 
@@ -348,7 +352,7 @@ cpdef tuple conditional_inner_join(
     Table left,
     Table right,
     Expression binary_predicate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a conditional inner join between two tables.
@@ -398,7 +402,7 @@ cpdef tuple conditional_left_join(
     Table left,
     Table right,
     Expression binary_predicate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a conditional left join between two tables.
@@ -448,7 +452,7 @@ cpdef tuple conditional_full_join(
     Table left,
     Table right,
     Expression binary_predicate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a conditional full join between two tables.
@@ -496,7 +500,7 @@ cpdef Column conditional_left_semi_join(
     Table left,
     Table right,
     Expression binary_predicate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a conditional left semi join between two tables.
@@ -542,7 +546,7 @@ cpdef Column conditional_left_anti_join(
     Table left,
     Table right,
     Expression binary_predicate,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a conditional left anti join between two tables.
@@ -591,7 +595,7 @@ cpdef tuple mixed_inner_join(
     Table right_conditional,
     Expression binary_predicate,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a mixed inner join between two tables.
@@ -655,7 +659,7 @@ cpdef tuple mixed_left_join(
     Table right_conditional,
     Expression binary_predicate,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a mixed left join between two tables.
@@ -719,7 +723,7 @@ cpdef tuple mixed_full_join(
     Table right_conditional,
     Expression binary_predicate,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a mixed full join between two tables.
@@ -783,7 +787,7 @@ cpdef Column mixed_left_semi_join(
     Table right_conditional,
     Expression binary_predicate,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a mixed left semi join between two tables.
@@ -841,7 +845,7 @@ cpdef Column mixed_left_anti_join(
     Table right_conditional,
     Expression binary_predicate,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a mixed left anti join between two tables.
@@ -910,7 +914,7 @@ cdef class FilteredJoin:
         Table right,
         null_equality compare_nulls=null_equality.EQUAL,
         double load_factor=0.5,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
     ):
         """
         Construct a filtered hash join object for subsequent probe calls.
@@ -944,7 +948,7 @@ cdef class FilteredJoin:
     def semi_join(
         self,
         Table left,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """
@@ -985,7 +989,7 @@ cdef class FilteredJoin:
     def anti_join(
         self,
         Table left,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None,
     ):
         """
diff --git a/python/pylibcudf/pylibcudf/json.pyx b/python/pylibcudf/pylibcudf/json.pyx
index 7d71f2a9dca4..ca1222f5385f 100644
--- a/python/pylibcudf/pylibcudf/json.pyx
+++ b/python/pylibcudf/pylibcudf/json.pyx
@@ -16,6 +16,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["GetJsonObjectOptions", "get_json_object"]
@@ -122,7 +126,7 @@ cpdef Column get_json_object(
     Column col,
     Scalar json_path,
     GetJsonObjectOptions options=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/labeling.pyx b/python/pylibcudf/pylibcudf/labeling.pyx
index 9d52e308f7c4..1dc6ba43c339 100644
--- a/python/pylibcudf/pylibcudf/labeling.pyx
+++ b/python/pylibcudf/pylibcudf/labeling.pyx
@@ -15,6 +15,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 
 from .column cimport Column
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["Inclusive", "label_bins"]
@@ -25,7 +29,7 @@ cpdef Column label_bins(
     inclusive left_inclusive,
     Column right_edges,
     inclusive right_inclusive,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Labels elements based on membership in the specified bins.
diff --git a/python/pylibcudf/pylibcudf/lists.pyx b/python/pylibcudf/pylibcudf/lists.pyx
index 71996340e5db..930cacec73dc 100644
--- a/python/pylibcudf/pylibcudf/lists.pyx
+++ b/python/pylibcudf/pylibcudf/lists.pyx
@@ -59,6 +59,10 @@ from .column cimport Column, ListsColumnView
 from .scalar cimport Scalar
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -88,7 +92,7 @@ __all__ = [
 cpdef Table explode_outer(
     Table input,
     size_type explode_column_idx,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Explode a column of lists into rows.
@@ -126,7 +130,7 @@ cpdef Table explode_outer(
 
 cpdef Column concatenate_rows(
     Table input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Concatenate multiple lists columns into a single lists column row-wise.
@@ -163,7 +167,7 @@ cpdef Column concatenate_rows(
 cpdef Column concatenate_list_elements(
     Column input,
     concatenate_null_policy null_policy,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Concatenate multiple lists on the same row into a single list.
@@ -200,7 +204,7 @@ cpdef Column concatenate_list_elements(
 cpdef Column contains(
     Column input,
     ColumnOrScalar search_key,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column of bool values indicating whether
@@ -257,7 +261,7 @@ cpdef Column contains(
 
 cpdef Column contains_nulls(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column of bool values indicating whether
@@ -297,7 +301,7 @@ cpdef Column index_of(
     Column input,
     ColumnOrScalar search_key,
     duplicate_find_option find_option,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column of index values indicating the position of a search
@@ -352,7 +356,7 @@ cpdef Column index_of(
 
 cpdef Column reverse(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Reverse the element order within each list of the input column.
@@ -388,7 +392,7 @@ cpdef Column segmented_gather(
     Column input,
     Column gather_map_list,
     out_of_bounds_policy bounds_policy=out_of_bounds_policy.DONT_CHECK,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column with elements gathered based on the indices in gather_map_list
@@ -443,7 +447,7 @@ cpdef Column segmented_gather(
 cpdef Column extract_list_element(
     Column input,
     ColumnOrSizeType index,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column of extracted list elements.
@@ -486,7 +490,7 @@ cpdef Column extract_list_element(
 
 cpdef Column count_elements(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Count the number of rows in each
@@ -525,7 +529,7 @@ cpdef Column sequences(
     Column starts,
     Column sizes,
     Column steps = None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a lists column in which each row contains a sequence of
@@ -583,7 +587,7 @@ cpdef Column sort_lists(
     order sort_order,
     null_order na_position,
     bool stable = False,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sort the elements within a list in each row of a list column.
@@ -641,7 +645,7 @@ cpdef Column difference_distinct(
     Column rhs,
     null_equality nulls_equal=null_equality.EQUAL,
     nan_equality nans_equal=nan_equality.ALL_EQUAL,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column of index values indicating the position of a search
@@ -692,7 +696,7 @@ cpdef Column have_overlap(
     Column rhs,
     null_equality nulls_equal=null_equality.EQUAL,
     nan_equality nans_equal=nan_equality.ALL_EQUAL,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Check if lists at each row of the given lists columns overlap.
@@ -742,7 +746,7 @@ cpdef Column intersect_distinct(
     Column rhs,
     null_equality nulls_equal=null_equality.EQUAL,
     nan_equality nans_equal=nan_equality.ALL_EQUAL,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a lists column of distinct elements common to two input lists columns.
@@ -792,7 +796,7 @@ cpdef Column union_distinct(
     Column rhs,
     null_equality nulls_equal=null_equality.EQUAL,
     nan_equality nans_equal=nan_equality.ALL_EQUAL,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a lists column of distinct elements found in
@@ -841,7 +845,7 @@ cpdef Column union_distinct(
 cpdef Column apply_boolean_mask(
     Column input,
     Column boolean_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters elements in each row of the input lists column using a boolean mask
@@ -885,7 +889,7 @@ cpdef Column apply_boolean_mask(
 cpdef Column apply_deletion_mask(
     Column input,
     Column deletion_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters elements in each row of the input lists column using a deletion mask.
@@ -928,7 +932,7 @@ cpdef Column distinct(
     Column input,
     null_equality nulls_equal,
     nan_equality nans_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a new list column without duplicate elements in each list.
diff --git a/python/pylibcudf/pylibcudf/merge.pyx b/python/pylibcudf/pylibcudf/merge.pyx
index 3c0cd93a342e..82c075b57d6a 100644
--- a/python/pylibcudf/pylibcudf/merge.pyx
+++ b/python/pylibcudf/pylibcudf/merge.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libcpp.memory cimport unique_ptr
@@ -16,14 +16,20 @@ from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
 from cuda.bindings.cyruntime cimport cudaStream_t
 
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.types import NullOrder, Order
+    from pylibcudf.typing import CudaStreamLike
+
 __all__ = ["merge"]
 
 cpdef Table merge (
-    list tables_to_merge,
-    list key_cols,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list tables_to_merge: list[Table],
+    list key_cols: list[int],
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Merge a set of sorted tables.
diff --git a/python/pylibcudf/pylibcudf/null_mask.pyx b/python/pylibcudf/pylibcudf/null_mask.pyx
index bdb92b66bb02..ef121704ea1a 100644
--- a/python/pylibcudf/pylibcudf/null_mask.pyx
+++ b/python/pylibcudf/pylibcudf/null_mask.pyx
@@ -1,5 +1,6 @@
 # SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
+from collections.abc import Sequence
 from libc.stdint cimport uintptr_t
 from libcpp.memory cimport make_unique
 from libcpp.pair cimport pair
@@ -21,6 +22,10 @@ from .span import is_span as py_is_span
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -43,7 +48,7 @@ cdef DeviceBuffer buffer_to_python(
 
 cpdef DeviceBuffer copy_bitmask(
     Column col,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Copies ``col``'s bitmask into a ``DeviceBuffer``.
@@ -81,7 +86,7 @@ cpdef DeviceBuffer copy_bitmask_from_bitmask(
     object bitmask,
     size_type begin_bit,
     size_type end_bit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Copies a portion of a bitmask into a ``DeviceBuffer``.
@@ -154,7 +159,7 @@ cpdef size_t bitmask_allocation_size_bytes(size_type number_of_bits):
 cpdef DeviceBuffer create_null_mask(
     size_type size,
     mask_state state = mask_state.UNINITIALIZED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Creates a ``DeviceBuffer`` for use as a null value indicator bitmask of a
@@ -192,7 +197,11 @@ cpdef DeviceBuffer create_null_mask(
     return buffer_to_python(move(db), _stream, mr)
 
 
-cpdef tuple bitmask_and(columns, object stream=None, DeviceMemoryResource mr=None):
+cpdef tuple bitmask_and(
+    columns: Sequence[Column],
+    object stream: CudaStreamLike | None = None,
+    DeviceMemoryResource mr=None,
+):
     """Performs bitwise AND of the bitmasks of a list of columns.
 
     For details, see :cpp:func:`bitmask_and`.
@@ -226,7 +235,11 @@ cpdef tuple bitmask_and(columns, object stream=None, DeviceMemoryResource mr=Non
     return buffer_to_python(move(c_result.first), _stream, mr), c_result.second
 
 
-cpdef tuple bitmask_or(columns, object stream=None, DeviceMemoryResource mr=None):
+cpdef tuple bitmask_or(
+    columns: Sequence[Column],
+    object stream: CudaStreamLike | None = None,
+    DeviceMemoryResource mr=None,
+):
     """Performs bitwise OR of the bitmasks of a list of columns.
 
     For details, see :cpp:func:`bitmask_or`.
@@ -262,7 +275,7 @@ cpdef size_type null_count(
     object bitmask,
     size_type start,
     size_type stop,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Given a validity bitmask, counts the number of null elements.
 
@@ -304,7 +317,7 @@ cpdef size_type index_of_first_set_bit(
     object bitmask,
     size_type start,
     size_type stop,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Given a validity bitmask, returns the index of the first valid element
     relative to ``start``.
diff --git a/python/pylibcudf/pylibcudf/nvtext/byte_pair_encode.pyx b/python/pylibcudf/pylibcudf/nvtext/byte_pair_encode.pyx
index 0db5a345f43c..8a20d0ebf41c 100644
--- a/python/pylibcudf/pylibcudf/nvtext/byte_pair_encode.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/byte_pair_encode.pyx
@@ -17,6 +17,10 @@ from pylibcudf.libcudf.scalar.scalar_factories cimport (
 )
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -31,7 +35,7 @@ cdef class BPEMergePairs:
     def __cinit__(
         self,
         Column merge_pairs,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
     ):
         cdef column_view c_pairs = merge_pairs.view()
@@ -49,7 +53,7 @@ cpdef Column byte_pair_encoding(
     Column input,
     BPEMergePairs merge_pairs,
     Scalar separator=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/deduplicate.pyx b/python/pylibcudf/pylibcudf/nvtext/deduplicate.pyx
index 1dbbe3e066e6..55e86c128eac 100644
--- a/python/pylibcudf/pylibcudf/nvtext/deduplicate.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/deduplicate.pyx
@@ -16,6 +16,10 @@ from pylibcudf.libcudf.nvtext.deduplicate cimport (
 )
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.librmm.device_buffer cimport device_buffer
 from rmm.pylibrmm.stream cimport Stream
@@ -43,7 +47,7 @@ cdef Column _column_from_suffix_array(
 
 
 cpdef Column build_suffix_array(
-    Column input, size_type min_width, object stream=None, DeviceMemoryResource mr=None
+    Column input, size_type min_width, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Builds a suffix array for the input strings column.
@@ -85,7 +89,7 @@ cpdef Column resolve_duplicates(
     Column input,
     Column indices,
     size_type min_width,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -133,7 +137,7 @@ cpdef Column resolve_duplicates_pair(
     Column input2,
     Column indices2,
     size_type min_width,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/edit_distance.pyx b/python/pylibcudf/pylibcudf/nvtext/edit_distance.pyx
index 06ae6a8366dc..03d45f82697a 100644
--- a/python/pylibcudf/pylibcudf/nvtext/edit_distance.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/edit_distance.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libcpp.memory cimport unique_ptr
@@ -14,6 +14,10 @@ from rmm.pylibrmm.stream cimport Stream
 
 from ..column cimport Column
 from ..utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["edit_distance"]
@@ -21,7 +25,7 @@ __all__ = ["edit_distance"]
 cpdef Column edit_distance(
     Column input,
     Column targets,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/generate_ngrams.pyx b/python/pylibcudf/pylibcudf/nvtext/generate_ngrams.pyx
index 6d70751a5a06..b4778f9077e4 100644
--- a/python/pylibcudf/pylibcudf/nvtext/generate_ngrams.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/generate_ngrams.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libc.stdint cimport uint32_t
@@ -16,6 +16,10 @@ from pylibcudf.libcudf.scalar.scalar cimport string_scalar
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -30,7 +34,7 @@ cpdef Column generate_ngrams(
     Column input,
     size_type ngrams,
     Scalar separator,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -75,7 +79,7 @@ cpdef Column generate_ngrams(
 cpdef Column generate_character_ngrams(
     Column input,
     size_type ngrams = 2,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -117,7 +121,7 @@ cpdef Column hash_character_ngrams(
     Column input,
     size_type ngrams,
     uint32_t seed,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/jaccard.pyx b/python/pylibcudf/pylibcudf/nvtext/jaccard.pyx
index 24a343e4508a..57129757f16d 100644
--- a/python/pylibcudf/pylibcudf/nvtext/jaccard.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/jaccard.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libcpp.memory cimport unique_ptr
@@ -11,6 +11,10 @@ from pylibcudf.libcudf.nvtext.jaccard cimport (
 )
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -21,7 +25,7 @@ cpdef Column jaccard_index(
     Column input1,
     Column input2,
     size_type width,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/minhash.pyx b/python/pylibcudf/pylibcudf/nvtext/minhash.pyx
index 035c4370acd2..92c9cbfd081f 100644
--- a/python/pylibcudf/pylibcudf/nvtext/minhash.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/minhash.pyx
@@ -15,6 +15,10 @@ from pylibcudf.libcudf.nvtext.minhash cimport (
 )
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -32,7 +36,7 @@ cpdef Column minhash(
     Column a,
     Column b,
     size_type width,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
@@ -86,7 +90,7 @@ cpdef Column minhash64(
     Column a,
     Column b,
     size_type width,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
@@ -142,7 +146,7 @@ cpdef Column minhash_ngrams(
     uint32_t seed,
     Column a,
     Column b,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
@@ -199,7 +203,7 @@ cpdef Column minhash64_ngrams(
     uint64_t seed,
     Column a,
     Column b,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/ngrams_tokenize.pyx b/python/pylibcudf/pylibcudf/nvtext/ngrams_tokenize.pyx
index 7dfba5c6b989..fccae239a514 100644
--- a/python/pylibcudf/pylibcudf/nvtext/ngrams_tokenize.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/ngrams_tokenize.pyx
@@ -14,6 +14,10 @@ from pylibcudf.libcudf.scalar.scalar cimport string_scalar
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -25,7 +29,7 @@ cpdef Column ngrams_tokenize(
     size_type ngrams,
     Scalar delimiter,
     Scalar separator,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/normalize.pyx b/python/pylibcudf/pylibcudf/nvtext/normalize.pyx
index 889792d67383..ec083549ccee 100644
--- a/python/pylibcudf/pylibcudf/nvtext/normalize.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/normalize.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.column.column cimport column
 from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.nvtext cimport normalize as cpp_normalize
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -29,7 +33,7 @@ cdef class CharacterNormalizer:
         self,
         bool do_lower_case,
         Column tokens,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
     ):
         cdef column_view c_tokens = tokens.view()
@@ -49,7 +53,7 @@ cdef class CharacterNormalizer:
     __hash__ = None
 
 cpdef Column normalize_spaces(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new strings column by normalizing the whitespace in
@@ -86,7 +90,7 @@ cpdef Column normalize_spaces(
 cpdef Column normalize_characters(
     Column input,
     CharacterNormalizer normalizer,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/replace.pyx b/python/pylibcudf/pylibcudf/nvtext/replace.pyx
index 70fec3d96640..3c5131cda0ce 100644
--- a/python/pylibcudf/pylibcudf/nvtext/replace.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/replace.pyx
@@ -18,6 +18,10 @@ from pylibcudf.libcudf.scalar.scalar_factories cimport (
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -29,7 +33,7 @@ cpdef Column replace_tokens(
     Column targets,
     Column replacements,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -84,7 +88,7 @@ cpdef Column filter_tokens(
     size_type min_token_length,
     Scalar replacement=None,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/stemmer.pyx b/python/pylibcudf/pylibcudf/nvtext/stemmer.pyx
index cb5c5f69cc23..44d20377dea1 100644
--- a/python/pylibcudf/pylibcudf/nvtext/stemmer.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/stemmer.pyx
@@ -15,6 +15,10 @@ from pylibcudf.libcudf.nvtext.stemmer cimport (
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.nvtext.stemmer cimport ColumnOrSize
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from pylibcudf.libcudf.nvtext.stemmer import letter_type as LetterType # no-cython-lint
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
@@ -27,7 +31,7 @@ cpdef Column is_letter(
     Column input,
     bool check_vowels,
     ColumnOrSize indices,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -77,7 +81,7 @@ cpdef Column is_letter(
 
 
 cpdef Column porter_stemmer_measure(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns the Porter Stemmer measurements of a strings column.
diff --git a/python/pylibcudf/pylibcudf/nvtext/tokenize.pyx b/python/pylibcudf/pylibcudf/nvtext/tokenize.pyx
index 2459760a9d4c..fa660fb80660 100644
--- a/python/pylibcudf/pylibcudf/nvtext/tokenize.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/tokenize.pyx
@@ -22,6 +22,10 @@ from pylibcudf.libcudf.scalar.scalar_factories cimport (
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -42,7 +46,7 @@ cdef class TokenizeVocabulary:
 
     For details, see :cpp:class:`cudf::nvtext::tokenize_vocabulary`.
     """
-    def __cinit__(self, Column vocab, object stream=None, DeviceMemoryResource mr=None):
+    def __cinit__(self, Column vocab, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
         cdef column_view c_vocab = vocab.view()
         cdef Stream _stream = _get_stream(stream)
         cdef cudaStream_t _cs = _stream.view().value()
@@ -55,7 +59,7 @@ cdef class TokenizeVocabulary:
 cpdef Column tokenize_scalar(
     Column input,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -100,7 +104,7 @@ cpdef Column tokenize_scalar(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column tokenize_column(
-    Column input, Column delimiters, object stream=None, DeviceMemoryResource mr=None
+    Column input, Column delimiters, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a single column of strings by tokenizing the input
@@ -142,7 +146,7 @@ cpdef Column tokenize_column(
 cpdef Column count_tokens_scalar(
     Column input,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -187,7 +191,7 @@ cpdef Column count_tokens_scalar(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column count_tokens_column(
-    Column input, Column delimiters, object stream=None, DeviceMemoryResource mr=None
+    Column input, Column delimiters, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns the number of tokens in each string of a strings column
@@ -227,7 +231,7 @@ cpdef Column count_tokens_column(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column character_tokenize(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a single column of strings by converting
@@ -261,7 +265,7 @@ cpdef Column detokenize(
     Column input,
     Column row_indices,
     Scalar separator=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -314,7 +318,7 @@ cpdef Column tokenize_with_vocabulary(
     TokenizeVocabulary vocabulary,
     Scalar delimiter,
     size_type default_id=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/nvtext/wordpiece_tokenize.pyx b/python/pylibcudf/pylibcudf/nvtext/wordpiece_tokenize.pyx
index 815139152b59..13b148e40c3b 100644
--- a/python/pylibcudf/pylibcudf/nvtext/wordpiece_tokenize.pyx
+++ b/python/pylibcudf/pylibcudf/nvtext/wordpiece_tokenize.pyx
@@ -13,6 +13,10 @@ from pylibcudf.libcudf.nvtext.wordpiece_tokenize cimport (
 )
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -30,7 +34,7 @@ cdef class WordPieceVocabulary:
     def __cinit__(
         self,
         Column vocab,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
     ):
         cdef column_view c_vocab = vocab.view()
@@ -48,7 +52,7 @@ cpdef Column wordpiece_tokenize(
     Column input,
     WordPieceVocabulary vocabulary,
     size_type max_words_per_row,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/partitioning.pyx b/python/pylibcudf/pylibcudf/partitioning.pyx
index 88f7f9de086e..bf2f1c001312 100644
--- a/python/pylibcudf/pylibcudf/partitioning.pyx
+++ b/python/pylibcudf/pylibcudf/partitioning.pyx
@@ -17,6 +17,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -32,7 +36,7 @@ cpdef tuple[Table, list] hash_partition(
     int num_partitions,
     cpp_partitioning.hash_id hash_function = cpp_partitioning.hash_id.HASH_MURMUR3,
     uint32_t seed = cpp_partitioning.DEFAULT_HASH_SEED,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -102,7 +106,7 @@ cpdef tuple[Table, list] partition(
     Table t,
     Column partition_map,
     int num_partitions,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -155,7 +159,7 @@ cpdef tuple[Table, list] round_robin_partition(
     Table input,
     int num_partitions,
     int start_partition=0,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/quantiles.pyx b/python/pylibcudf/pylibcudf/quantiles.pyx
index 6f28720836ef..0651f3fea22b 100644
--- a/python/pylibcudf/pylibcudf/quantiles.pyx
+++ b/python/pylibcudf/pylibcudf/quantiles.pyx
@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
+from collections.abc import Iterable
 from libcpp cimport bool
 from libcpp.memory cimport unique_ptr
 from libcpp.utility cimport move
@@ -21,17 +22,21 @@ from .column cimport Column
 from .table cimport Table
 from .types cimport interpolation
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["quantile", "quantiles"]
 
 cpdef Column quantile(
     Column input,
-    vector[double] q,
+    vector[double] q: Iterable[float],
     interpolation interp = interpolation.LINEAR,
     Column ordered_indices = None,
     bool exact=True,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes quantiles with interpolation.
@@ -97,12 +102,12 @@ cpdef Column quantile(
 
 cpdef Table quantiles(
     Table input,
-    vector[double] q,
+    vector[double] q: Iterable[float],
     interpolation interp = interpolation.NEAREST,
     sorted is_input_sorted = sorted.NO,
     list column_order = None,
     list null_precedence = None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes row quantiles with interpolation.
diff --git a/python/pylibcudf/pylibcudf/reduce.pyx b/python/pylibcudf/pylibcudf/reduce.pyx
index b2bdfa720872..4e6d0aec5607 100644
--- a/python/pylibcudf/pylibcudf/reduce.pyx
+++ b/python/pylibcudf/pylibcudf/reduce.pyx
@@ -31,6 +31,10 @@ from .column cimport Column
 from .scalar cimport Scalar
 from .types cimport DataType
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from pylibcudf.libcudf.reduce import scan_type as ScanType  # no-cython-lint
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -50,7 +54,7 @@ cpdef Scalar reduce(
     Aggregation agg,
     DataType data_type,
     Scalar init=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a reduction on a column
@@ -111,7 +115,7 @@ cpdef Column scan(
     Column col,
     Aggregation agg,
     scan_type inclusive,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a scan on a column
@@ -156,7 +160,7 @@ cpdef Column scan(
     return Column.from_libcudf(move(result), _stream, mr)
 
 
-cpdef tuple minmax(Column col, object stream=None, DeviceMemoryResource mr=None):
+cpdef tuple minmax(Column col, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Compute the minimum and maximum of a column
 
     For details, see ``cudf::minmax`` documentation.
@@ -215,7 +219,7 @@ cpdef size_type unique_count(
     Column source,
     null_policy null_handling,
     nan_policy nan_handling,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Returns the number of unique consecutive elements in the input column.
 
@@ -255,7 +259,7 @@ cpdef size_type distinct_count(
     Column source,
     null_policy null_handling,
     nan_policy nan_handling,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Returns the number of distinct elements in the input column.
 
@@ -289,7 +293,7 @@ cpdef size_type distinct_count(
 cpdef size_type unique_count_table(
     Table source,
     null_equality nulls_equal,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Returns the number of unique consecutive rows in the input table.
 
@@ -323,7 +327,7 @@ cpdef size_type unique_count_table(
 cpdef size_type distinct_count_table(
     Table source,
     null_equality nulls_equal,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """Returns the number of distinct rows in the input table.
 
diff --git a/python/pylibcudf/pylibcudf/replace.pyx b/python/pylibcudf/pylibcudf/replace.pyx
index e411ae645062..604a3cefa40c 100644
--- a/python/pylibcudf/pylibcudf/replace.pyx
+++ b/python/pylibcudf/pylibcudf/replace.pyx
@@ -19,6 +19,10 @@ from pylibcudf.libcudf.replace import \
 from .column cimport Column
 from .scalar cimport Scalar
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -33,7 +37,7 @@ __all__ = [
 cpdef Column replace_nulls(
     Column source_column,
     ReplacementType replacement,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Replace nulls in source_column.
@@ -130,7 +134,7 @@ cpdef Column find_and_replace_all(
     Column source_column,
     Column values_to_replace,
     Column replacement_values,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Replace all occurrences of values_to_replace with replacement_values.
@@ -182,7 +186,7 @@ cpdef Column clamp(
     Scalar hi,
     Scalar lo_replace=None,
     Scalar hi_replace=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Clamp the values in source_column to the range [lo, hi].
@@ -248,7 +252,7 @@ cpdef Column clamp(
 cpdef Column normalize_nans_and_zeros(
     Column source_column,
     bool inplace=False,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Normalize NaNs and zeros in source_column.
diff --git a/python/pylibcudf/pylibcudf/reshape.pyx b/python/pylibcudf/pylibcudf/reshape.pyx
index b1b60f4444dd..211bda522ab4 100644
--- a/python/pylibcudf/pylibcudf/reshape.pyx
+++ b/python/pylibcudf/pylibcudf/reshape.pyx
@@ -25,12 +25,16 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["interleave_columns", "tile", "table_to_array"]
 
 cpdef Column interleave_columns(
-    Table source_table, object stream=None, DeviceMemoryResource mr=None
+    Table source_table, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Interleave columns of a table into a single column.
 
@@ -73,7 +77,7 @@ cpdef Column interleave_columns(
 cpdef Table tile(
     Table source_table,
     size_type count,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Repeats the rows from input table count times to form a new table.
@@ -114,7 +118,7 @@ cpdef void table_to_array(
     Table input_table,
     uintptr_t ptr,
     size_t size,
-    object stream=None
+    object stream: CudaStreamLike | None = None
 ):
     """
     Copy a table into a preallocated column-major device array.
diff --git a/python/pylibcudf/pylibcudf/rolling.pyx b/python/pylibcudf/pylibcudf/rolling.pyx
index d12a40592983..b6e5e12f2635 100644
--- a/python/pylibcudf/pylibcudf/rolling.pyx
+++ b/python/pylibcudf/pylibcudf/rolling.pyx
@@ -22,6 +22,10 @@ from .column cimport Column
 from .scalar cimport Scalar
 from .types cimport DataType
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 
@@ -127,8 +131,8 @@ cpdef Table grouped_range_rolling_window(
     null_order null_order,
     PrecedingRangeWindowType preceding,
     FollowingRangeWindowType following,
-    list requests,
-    object stream=None,
+    list requests: list[RollingRequest],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -194,7 +198,7 @@ cpdef Column rolling_window(
     WindowType following_window,
     size_type min_periods,
     Aggregation agg,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a rolling window operation on a column
@@ -290,7 +294,7 @@ cpdef tuple make_range_windows(
     null_order null_order,
     PrecedingRangeWindowType preceding,
     FollowingRangeWindowType following,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/round.pyx b/python/pylibcudf/pylibcudf/round.pyx
index a3de7add7533..7a6585c36a08 100644
--- a/python/pylibcudf/pylibcudf/round.pyx
+++ b/python/pylibcudf/pylibcudf/round.pyx
@@ -20,6 +20,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 
 from .column cimport Column
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["RoundingMethod", "round"]
@@ -28,7 +32,7 @@ cpdef Column round(
     Column source,
     int32_t decimal_places = 0,
     rounding_method round_method = rounding_method.HALF_UP,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Rounds all the values in a column to the specified number of decimal places.
@@ -81,7 +85,7 @@ cpdef Column round_decimal(
     Column source,
     int32_t decimal_places = 0,
     rounding_method round_method = rounding_method.HALF_UP,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """Rounds all the values in a column to the specified number of decimal places.
diff --git a/python/pylibcudf/pylibcudf/scalar.pyx b/python/pylibcudf/pylibcudf/scalar.pyx
index 1b80a62bed96..2a663104a8d6 100644
--- a/python/pylibcudf/pylibcudf/scalar.pyx
+++ b/python/pylibcudf/pylibcudf/scalar.pyx
@@ -64,6 +64,10 @@ from .column cimport Column
 from .traits cimport is_floating_point
 from .types cimport DataType
 from .utils cimport _get_memory_resource, _get_stream
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from functools import singledispatch
 from ._interop_helpers import ArrowLike, ColumnMetadata
 
@@ -153,7 +157,7 @@ cdef class Scalar:
         """The type of data in the column."""
         return self._data_type
 
-    cpdef bool is_valid(self, object stream = None):
+    cpdef bool is_valid(self, object stream: CudaStreamLike | None = None):
         """True if the scalar is valid, false if not"""
         cdef Stream _stream = _get_stream(stream)
         cdef cudaStream_t _cs = _stream.view().value()
diff --git a/python/pylibcudf/pylibcudf/search.pyx b/python/pylibcudf/pylibcudf/search.pyx
index d67229729d7b..d76d427c0647 100644
--- a/python/pylibcudf/pylibcudf/search.pyx
+++ b/python/pylibcudf/pylibcudf/search.pyx
@@ -17,14 +17,20 @@ from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
 from cuda.bindings.cyruntime cimport cudaStream_t
 
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.types import NullOrder, Order
+    from pylibcudf.typing import CudaStreamLike
+
 __all__ = ["contains", "lower_bound", "upper_bound"]
 
 cpdef Column lower_bound(
     Table haystack,
     Table needles,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Find smallest indices in haystack where needles may be inserted to retain order.
@@ -76,9 +82,9 @@ cpdef Column lower_bound(
 cpdef Column upper_bound(
     Table haystack,
     Table needles,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Find largest indices in haystack where needles may be inserted to retain order.
@@ -128,7 +134,7 @@ cpdef Column upper_bound(
 
 
 cpdef Column contains(
-    Column haystack, Column needles, object stream=None, DeviceMemoryResource mr=None
+    Column haystack, Column needles, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Check whether needles are present in haystack.
 
diff --git a/python/pylibcudf/pylibcudf/sorting.pyx b/python/pylibcudf/pylibcudf/sorting.pyx
index eac72a4cdda6..fa6bc5bda3b5 100644
--- a/python/pylibcudf/pylibcudf/sorting.pyx
+++ b/python/pylibcudf/pylibcudf/sorting.pyx
@@ -19,6 +19,12 @@ from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
 from cuda.bindings.cyruntime cimport cudaStream_t
 
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.types import NullOrder, Order
+    from pylibcudf.typing import CudaStreamLike
+
 __all__ = [
     "is_sorted",
     "rank",
@@ -36,9 +42,9 @@ __all__ = [
 
 cpdef Column sorted_order(
     Table source_table,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the row indices required to sort the table.
@@ -81,9 +87,9 @@ cpdef Column sorted_order(
 
 cpdef Column stable_sorted_order(
     Table source_table,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the row indices required to sort the table,
@@ -132,7 +138,7 @@ cpdef Column rank(
     null_policy null_handling,
     null_order null_precedence,
     bool percentage,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Computes the rank of each element in the column.
@@ -181,7 +187,10 @@ cpdef Column rank(
 
 
 cpdef bool is_sorted(
-    Table tbl, list column_order, list null_precedence, object stream=None
+    Table tbl,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
 ):
     """Checks if the table is sorted.
 
@@ -223,9 +232,9 @@ cpdef Table segmented_sort_by_key(
     Table values,
     Table keys,
     Column segment_offsets,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table by key, within segments.
@@ -278,9 +287,9 @@ cpdef Table stable_segmented_sort_by_key(
     Table values,
     Table keys,
     Column segment_offsets,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table by key preserving order of equal elements,
@@ -333,9 +342,9 @@ cpdef Table stable_segmented_sort_by_key(
 cpdef Table sort_by_key(
     Table values,
     Table keys,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table by key.
@@ -383,9 +392,9 @@ cpdef Table sort_by_key(
 cpdef Table stable_sort_by_key(
     Table values,
     Table keys,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table by key preserving order of equal elements.
@@ -432,9 +441,9 @@ cpdef Table stable_sort_by_key(
 
 cpdef Table sort(
     Table source_table,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table.
@@ -477,9 +486,9 @@ cpdef Table sort(
 
 cpdef Table stable_sort(
     Table source_table,
-    list column_order,
-    list null_precedence,
-    object stream=None,
+    list column_order: list[Order],
+    list null_precedence: list[NullOrder],
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Sorts the table preserving order of equal elements.
@@ -524,7 +533,7 @@ cpdef Column top_k(
     Column col,
     size_type k,
     order sort_order = order.DESCENDING,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -569,7 +578,7 @@ cpdef Column top_k_order(
     Column col,
     size_type k,
     order sort_order = order.DESCENDING,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/stream_compaction.pyx b/python/pylibcudf/pylibcudf/stream_compaction.pyx
index 457c82f63a7e..9d64484c9605 100644
--- a/python/pylibcudf/pylibcudf/stream_compaction.pyx
+++ b/python/pylibcudf/pylibcudf/stream_compaction.pyx
@@ -26,6 +26,10 @@ from .column cimport Column
 from .expressions cimport Expression
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -43,9 +47,9 @@ __all__ = [
 
 cpdef Table drop_nulls(
     Table source_table,
-    list keys,
+    list keys: list[int],
     size_type keep_threshold,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters out rows from the input table based on the presence of nulls.
@@ -83,9 +87,9 @@ cpdef Table drop_nulls(
 
 cpdef Table drop_nans(
     Table source_table,
-    list keys,
+    list keys: list[int],
     size_type keep_threshold,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters out rows from the input table based on the presence of NaNs.
@@ -124,7 +128,7 @@ cpdef Table drop_nans(
 cpdef Table apply_boolean_mask(
     Table source_table,
     Column boolean_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters out rows from the input table based on a boolean mask.
@@ -161,7 +165,7 @@ cpdef Table apply_boolean_mask(
 cpdef Table apply_deletion_mask(
     Table source_table,
     Column deletion_mask,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters out rows from the input table using a deletion mask.
@@ -197,10 +201,10 @@ cpdef Table apply_deletion_mask(
 
 cpdef Table unique(
     Table input,
-    list keys,
+    list keys: list[int],
     duplicate_keep_option keep,
     null_equality nulls_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filter duplicate consecutive rows from the input table.
@@ -246,11 +250,11 @@ cpdef Table unique(
 
 cpdef Table distinct(
     Table input,
-    list keys,
+    list keys: list[int],
     duplicate_keep_option keep,
     null_equality nulls_equal,
     nan_equality nans_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Get the distinct rows from the input table.
@@ -297,7 +301,7 @@ cpdef Column distinct_indices(
     duplicate_keep_option keep,
     null_equality nulls_equal,
     nan_equality nans_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Get the indices of the distinct rows from the input table.
@@ -336,11 +340,11 @@ cpdef Column distinct_indices(
 
 cpdef Table stable_distinct(
     Table input,
-    list keys,
+    list keys: list[int],
     duplicate_keep_option keep,
     null_equality nulls_equal,
     nan_equality nans_equal,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Get the distinct rows from the input table, preserving input order.
@@ -386,7 +390,7 @@ cpdef Table filter(
     Table predicate_table,
     Expression predicate_expr,
     Table filter_table,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Filters a table using a predicate expression.
diff --git a/python/pylibcudf/pylibcudf/strings/attributes.pyx b/python/pylibcudf/pylibcudf/strings/attributes.pyx
index ead494d047f3..5d6597a1a3f7 100644
--- a/python/pylibcudf/pylibcudf/strings/attributes.pyx
+++ b/python/pylibcudf/pylibcudf/strings/attributes.pyx
@@ -8,6 +8,10 @@ from pylibcudf.libcudf.column.column cimport column
 from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport attributes as cpp_attributes
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -15,7 +19,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["code_points", "count_bytes", "count_characters"]
 
 cpdef Column count_characters(
-    Column source_strings, object stream=None, DeviceMemoryResource mr=None
+    Column source_strings, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a column containing character lengths of each string
@@ -47,7 +51,7 @@ cpdef Column count_characters(
 
 
 cpdef Column count_bytes(
-    Column source_strings, object stream=None, DeviceMemoryResource mr=None
+    Column source_strings, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a column containing byte lengths of each string
@@ -79,7 +83,7 @@ cpdef Column count_bytes(
 
 
 cpdef Column code_points(
-    Column source_strings, object stream=None, DeviceMemoryResource mr=None
+    Column source_strings, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Creates a numeric column with code point values (integers)
diff --git a/python/pylibcudf/pylibcudf/strings/capitalize.pyx b/python/pylibcudf/pylibcudf/strings/capitalize.pyx
index 2c5683980191..0c07c6d6baa1 100644
--- a/python/pylibcudf/pylibcudf/strings/capitalize.pyx
+++ b/python/pylibcudf/pylibcudf/strings/capitalize.pyx
@@ -14,6 +14,10 @@ from pylibcudf.libcudf.strings cimport capitalize as cpp_capitalize
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.strings.char_types cimport string_character_types
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -25,7 +29,7 @@ __all__ = ["capitalize", "is_title", "title"]
 cpdef Column capitalize(
     Column input,
     Scalar delimiters=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
     # TODO: default scalar values
     # https://github.com/rapidsai/cudf/issues/15505
@@ -76,7 +80,7 @@ cpdef Column capitalize(
 cpdef Column title(
     Column input,
     string_character_types sequence_type=string_character_types.ALPHA,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Modifies first character of each word to upper-case and lower-cases
@@ -109,7 +113,7 @@ cpdef Column title(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef Column is_title(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_title(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Checks if the strings in the input column are title formatted.
 
     For details, see :cpp:func:`is_title`.
diff --git a/python/pylibcudf/pylibcudf/strings/case.pyx b/python/pylibcudf/pylibcudf/strings/case.pyx
index 5a122faacb16..111af660b230 100644
--- a/python/pylibcudf/pylibcudf/strings/case.pyx
+++ b/python/pylibcudf/pylibcudf/strings/case.pyx
@@ -8,13 +8,17 @@ from pylibcudf.libcudf.column.column cimport column
 from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport case as cpp_case
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["swapcase", "to_lower", "to_upper"]
 
-cpdef Column to_lower(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column to_lower(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Returns a column of lowercased strings.
 
     For details, see :cpp:func:`to_lower`.
@@ -43,7 +47,7 @@ cpdef Column to_lower(Column input, object stream=None, DeviceMemoryResource mr=
 
     return Column.from_libcudf(move(c_result), _stream, mr)
 
-cpdef Column to_upper(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column to_upper(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Returns a column of uppercased strings.
 
     For details, see :cpp:func:`to_upper`.
@@ -72,7 +76,7 @@ cpdef Column to_upper(Column input, object stream=None, DeviceMemoryResource mr=
 
     return Column.from_libcudf(move(c_result), _stream, mr)
 
-cpdef Column swapcase(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column swapcase(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Returns a column of strings where the lowercase characters
     are converted to uppercase and the uppercase characters
     are converted to lowercase.
diff --git a/python/pylibcudf/pylibcudf/strings/char_types.pyx b/python/pylibcudf/pylibcudf/strings/char_types.pyx
index d7e155f548a5..10e03fee8552 100644
--- a/python/pylibcudf/pylibcudf/strings/char_types.pyx
+++ b/python/pylibcudf/pylibcudf/strings/char_types.pyx
@@ -11,6 +11,10 @@ from pylibcudf.libcudf.strings cimport char_types as cpp_char_types
 from pylibcudf.libcudf.strings.char_types cimport string_character_types
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -29,7 +33,7 @@ cpdef Column all_characters_of_type(
     Column source_strings,
     string_character_types types,
     string_character_types verify_types,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -72,7 +76,7 @@ cpdef Column filter_characters_of_type(
     string_character_types types_to_remove,
     Scalar replacement,
     string_character_types types_to_keep,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/combine.pyx b/python/pylibcudf/pylibcudf/strings/combine.pyx
index f147374df11d..a2fabe7352c0 100644
--- a/python/pylibcudf/pylibcudf/strings/combine.pyx
+++ b/python/pylibcudf/pylibcudf/strings/combine.pyx
@@ -14,6 +14,10 @@ from pylibcudf.scalar cimport Scalar
 from pylibcudf.table cimport Table
 from pylibcudf.libcudf.table.table_view cimport table_view
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -38,7 +42,7 @@ cpdef Column concatenate(
     Scalar narep=None,
     Scalar col_narep=None,
     separator_on_nulls separate_nulls=separator_on_nulls.YES,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -134,7 +138,7 @@ cpdef Column join_strings(
     Column input,
     Scalar separator,
     Scalar narep,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -190,7 +194,7 @@ cpdef Column join_list_elements(
     Scalar string_narep,
     separator_on_nulls separate_nulls,
     output_if_empty_list empty_list_policy,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/contains.pyx b/python/pylibcudf/pylibcudf/strings/contains.pyx
index 2e2e89de2795..a38a46a2ad92 100644
--- a/python/pylibcudf/pylibcudf/strings/contains.pyx
+++ b/python/pylibcudf/pylibcudf/strings/contains.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport contains as cpp_contains
 from pylibcudf.strings.regex_program cimport RegexProgram
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -19,7 +23,7 @@ __all__ = ["contains_re", "count_re", "like", "matches_re"]
 cpdef Column contains_re(
     Column input,
     RegexProgram prog,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Returns a boolean column identifying rows which match the given
@@ -61,7 +65,7 @@ cpdef Column contains_re(
 cpdef Column count_re(
     Column input,
     RegexProgram prog,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Returns the number of times the given regex_program's pattern
@@ -101,7 +105,7 @@ cpdef Column count_re(
 cpdef Column matches_re(
     Column input,
     RegexProgram prog,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Returns a boolean column identifying rows which
@@ -143,7 +147,7 @@ cpdef Column like(
     Column input,
     str pattern,
     str escape_character=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_booleans.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_booleans.pyx
index 2a4f77f4c0e8..020fdb066ae3 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_booleans.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_booleans.pyx
@@ -11,6 +11,10 @@ from pylibcudf.libcudf.strings.convert cimport (
 )
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from cython.operator import dereference
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
@@ -20,7 +24,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["from_booleans", "to_booleans"]
 
 cpdef Column to_booleans(
-    Column input, Scalar true_string, object stream=None, DeviceMemoryResource mr=None
+    Column input, Scalar true_string, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new bool column by parsing boolean values from the strings
@@ -67,7 +71,7 @@ cpdef Column from_booleans(
     Column booleans,
     Scalar true_string,
     Scalar false_string,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_datetime.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_datetime.pyx
index 9aab9ae06dce..49569055b503 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_datetime.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_datetime.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings.convert cimport (
     convert_datetime as cpp_convert_datetime,
 )
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -22,7 +26,7 @@ cpdef Column to_timestamps(
     Column input,
     DataType timestamp_type,
     str format,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -71,7 +75,7 @@ cpdef Column from_timestamps(
     Column timestamps,
     str format,
     Column input_strings_names,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -120,7 +124,7 @@ cpdef Column from_timestamps(
 cpdef Column is_timestamp(
     Column input,
     str format,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_durations.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_durations.pyx
index 8068d80a8e92..afe319a3ffdc 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_durations.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_durations.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings.convert cimport (
     convert_durations as cpp_convert_durations,
 )
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -22,7 +26,7 @@ cpdef Column to_durations(
     Column input,
     DataType duration_type,
     str format,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
@@ -71,7 +75,7 @@ cpdef Column to_durations(
 cpdef Column from_durations(
     Column durations,
     str format=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_fixed_point.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_fixed_point.pyx
index abc7d7a7848a..7ee321ffd646 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_fixed_point.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_fixed_point.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings.convert cimport (
 )
 from pylibcudf.types cimport DataType, type_id
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -18,7 +22,7 @@ __all__ = ["from_fixed_point", "is_fixed_point", "to_fixed_point"]
 
 
 cpdef Column to_fixed_point(
-    Column input, DataType output_type, object stream=None, DeviceMemoryResource mr=None
+    Column input, DataType output_type, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new fixed-point column parsing decimal values from the
@@ -59,7 +63,7 @@ cpdef Column to_fixed_point(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column from_fixed_point(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new strings column converting the fixed-point values
@@ -96,7 +100,7 @@ cpdef Column from_fixed_point(
 cpdef Column is_fixed_point(
     Column input,
     DataType decimal_type=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_floats.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_floats.pyx
index cd48c98e9d92..c73db3eda134 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_floats.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_floats.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings.convert cimport (
 )
 from pylibcudf.types cimport DataType
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -19,7 +23,7 @@ __all__ = ["from_floats", "is_float", "to_floats"]
 cpdef Column to_floats(
     Column strings,
     DataType output_type,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -62,7 +66,7 @@ cpdef Column to_floats(
 
 
 cpdef Column from_floats(
-    Column floats, object stream=None, DeviceMemoryResource mr=None
+    Column floats, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new strings column converting the float values from the
@@ -97,7 +101,7 @@ cpdef Column from_floats(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef Column is_float(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_float(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Returns a boolean column identifying strings in which all
     characters are valid for conversion to floats.
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_integers.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_integers.pyx
index 95db3b8829b5..f0c78e852e8e 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_integers.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_integers.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings.convert cimport (
 )
 from pylibcudf.types cimport DataType
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -24,7 +28,7 @@ __all__ = [
 ]
 
 cpdef Column to_integers(
-    Column input, DataType output_type, object stream=None, DeviceMemoryResource mr=None
+    Column input, DataType output_type, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new integer numeric column parsing integer values from the
@@ -68,7 +72,7 @@ cpdef Column to_integers(
 
 
 cpdef Column from_integers(
-    Column integers, object stream=None, DeviceMemoryResource mr=None
+    Column integers, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new strings column converting the integer values from the
@@ -110,7 +114,7 @@ cpdef Column from_integers(
 cpdef Column is_integer(
     Column input,
     DataType int_type=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -167,7 +171,7 @@ cpdef Column is_integer(
 
 
 cpdef Column hex_to_integers(
-    Column input, DataType output_type, object stream=None, DeviceMemoryResource mr=None
+    Column input, DataType output_type, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new integer numeric column parsing hexadecimal values
@@ -210,7 +214,7 @@ cpdef Column hex_to_integers(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef Column is_hex(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_hex(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Returns a boolean column identifying strings in which all
     characters are valid for conversion to integers from hex.
@@ -249,7 +253,7 @@ cpdef Column is_hex(Column input, object stream=None, DeviceMemoryResource mr=No
 
 
 cpdef Column integers_to_hex(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a new strings column converting integer columns to hexadecimal
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_ipv4.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_ipv4.pyx
index f0a262192b8b..3f3ada06ccb2 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_ipv4.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_ipv4.pyx
@@ -7,6 +7,10 @@ from pylibcudf.column cimport Column
 from pylibcudf.libcudf.column.column cimport column, column_view
 from pylibcudf.libcudf.strings.convert cimport convert_ipv4 as cpp_convert_ipv4
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -14,7 +18,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["integers_to_ipv4", "ipv4_to_integers", "is_ipv4"]
 
 cpdef Column ipv4_to_integers(
-    Column input, object stream=None, DeviceMemoryResource mr=None
+    Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Converts IPv4 addresses into integers.
@@ -49,7 +53,7 @@ cpdef Column ipv4_to_integers(
 
 
 cpdef Column integers_to_ipv4(
-    Column integers, object stream=None, DeviceMemoryResource mr=None
+    Column integers, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Converts integers into IPv4 addresses as strings.
@@ -83,7 +87,7 @@ cpdef Column integers_to_ipv4(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef Column is_ipv4(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_ipv4(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Returns a boolean column identifying strings in which all
     characters are valid for conversion to integers from IPv4 format.
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_lists.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_lists.pyx
index 903b83a9ea92..af86e771232b 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_lists.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_lists.pyx
@@ -16,6 +16,10 @@ from pylibcudf.libcudf.strings.convert cimport (
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.types cimport type_id
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -28,7 +32,7 @@ cpdef Column format_list_column(
     Column input,
     Scalar na_rep=None,
     Column separators=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/convert/convert_urls.pyx b/python/pylibcudf/pylibcudf/strings/convert/convert_urls.pyx
index 5c9d4bf17a79..0cd2ebb96d5d 100644
--- a/python/pylibcudf/pylibcudf/strings/convert/convert_urls.pyx
+++ b/python/pylibcudf/pylibcudf/strings/convert/convert_urls.pyx
@@ -7,6 +7,10 @@ from pylibcudf.column cimport Column
 from pylibcudf.libcudf.column.column cimport column, column_view
 from pylibcudf.libcudf.strings.convert cimport convert_urls as cpp_convert_urls
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
@@ -14,7 +18,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["url_decode", "url_encode"]
 
-cpdef Column url_encode(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column url_encode(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Encodes each string using URL encoding.
 
@@ -47,7 +51,7 @@ cpdef Column url_encode(Column input, object stream=None, DeviceMemoryResource m
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 
-cpdef Column url_decode(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column url_decode(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """
     Decodes each string using URL encoding.
 
diff --git a/python/pylibcudf/pylibcudf/strings/extract.pyx b/python/pylibcudf/pylibcudf/strings/extract.pyx
index 3dee27693a37..3ca69488d476 100644
--- a/python/pylibcudf/pylibcudf/strings/extract.pyx
+++ b/python/pylibcudf/pylibcudf/strings/extract.pyx
@@ -12,6 +12,10 @@ from pylibcudf.strings.regex_program cimport RegexProgram
 from pylibcudf.table cimport Table
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -19,7 +23,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["extract", "extract_all_record", "extract_single"]
 
 cpdef Table extract(
-    Column input, RegexProgram prog, object stream=None, DeviceMemoryResource mr=None
+    Column input, RegexProgram prog, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a table of strings columns where each column
@@ -59,7 +63,7 @@ cpdef Table extract(
 
 
 cpdef Column extract_all_record(
-    Column input, RegexProgram prog, object stream=None, DeviceMemoryResource mr=None
+    Column input, RegexProgram prog, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a lists column of strings where each string column
@@ -102,7 +106,7 @@ cpdef Column extract_single(
     Column input,
     RegexProgram prog,
     size_type group,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/find.pyx b/python/pylibcudf/pylibcudf/strings/find.pyx
index d9c9e81dea7d..bcd9f86c73d7 100644
--- a/python/pylibcudf/pylibcudf/strings/find.pyx
+++ b/python/pylibcudf/pylibcudf/strings/find.pyx
@@ -9,6 +9,10 @@ from pylibcudf.libcudf.strings cimport find as cpp_find
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -24,7 +28,7 @@ cpdef Column find(
     ColumnOrScalar target,
     size_type start=0,
     size_type stop=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Returns a column of character position values where the target string is
@@ -98,7 +102,7 @@ cpdef Column rfind(
     Scalar target,
     size_type start=0,
     size_type stop=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -146,7 +150,7 @@ cpdef Column rfind(
 cpdef Column contains(
     Column input,
     ColumnOrScalar target,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -212,7 +216,7 @@ cpdef Column contains(
 cpdef Column starts_with(
     Column input,
     ColumnOrScalar target,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -278,7 +282,7 @@ cpdef Column starts_with(
 cpdef Column ends_with(
     Column input,
     ColumnOrScalar target,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/find_multiple.pyx b/python/pylibcudf/pylibcudf/strings/find_multiple.pyx
index 3ac87788cd6f..611f8347613c 100644
--- a/python/pylibcudf/pylibcudf/strings/find_multiple.pyx
+++ b/python/pylibcudf/pylibcudf/strings/find_multiple.pyx
@@ -10,6 +10,10 @@ from pylibcudf.libcudf.strings cimport find_multiple as cpp_find_multiple
 from pylibcudf.libcudf.table.table cimport table
 from pylibcudf.table cimport Table
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -19,7 +23,7 @@ __all__ = ["find_multiple", "contains_multiple"]
 cpdef Column find_multiple(
     Column input,
     Column targets,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -62,7 +66,7 @@ cpdef Column find_multiple(
 cpdef Table contains_multiple(
     Column input,
     Column targets,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/findall.pyx b/python/pylibcudf/pylibcudf/strings/findall.pyx
index 17099870f3aa..214f2c61b4f5 100644
--- a/python/pylibcudf/pylibcudf/strings/findall.pyx
+++ b/python/pylibcudf/pylibcudf/strings/findall.pyx
@@ -9,6 +9,10 @@ from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport findall as cpp_findall
 from pylibcudf.strings.regex_program cimport RegexProgram
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -16,7 +20,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["findall", "find_re"]
 
 cpdef Column findall(
-    Column input, RegexProgram pattern, object stream=None, DeviceMemoryResource mr=None
+    Column input, RegexProgram pattern, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns a lists column of strings for each matching occurrence using
@@ -55,7 +59,7 @@ cpdef Column findall(
 
 
 cpdef Column find_re(
-    Column input, RegexProgram pattern, object stream=None, DeviceMemoryResource mr=None
+    Column input, RegexProgram pattern, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Returns character positions where the pattern first matches
diff --git a/python/pylibcudf/pylibcudf/strings/padding.pyx b/python/pylibcudf/pylibcudf/strings/padding.pyx
index 56949d84eb5f..8aafea761109 100644
--- a/python/pylibcudf/pylibcudf/strings/padding.pyx
+++ b/python/pylibcudf/pylibcudf/strings/padding.pyx
@@ -9,6 +9,10 @@ from pylibcudf.libcudf.strings cimport padding as cpp_padding
 from pylibcudf.libcudf.strings.side_type cimport side_type
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -20,7 +24,7 @@ cpdef Column pad(
     size_type width,
     side_type side,
     str fill_char,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -65,7 +69,7 @@ cpdef Column pad(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column zfill(
-    Column input, size_type width, object stream=None, DeviceMemoryResource mr=None
+    Column input, size_type width, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Add '0' as padding to the left of each string.
@@ -102,7 +106,7 @@ cpdef Column zfill(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column zfill_by_widths(
-    Column input, Column widths, object stream=None, DeviceMemoryResource mr=None
+    Column input, Column widths, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Add '0' as padding to the left of each string.
diff --git a/python/pylibcudf/pylibcudf/strings/repeat.pyx b/python/pylibcudf/pylibcudf/strings/repeat.pyx
index 3f3fd1f0310a..fb5f12415d92 100644
--- a/python/pylibcudf/pylibcudf/strings/repeat.pyx
+++ b/python/pylibcudf/pylibcudf/strings/repeat.pyx
@@ -12,6 +12,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
 from ..utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["repeat_strings"]
@@ -19,7 +23,7 @@ __all__ = ["repeat_strings"]
 cpdef Column repeat_strings(
     Column input,
     ColumnorSizeType repeat_times,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/replace.pyx b/python/pylibcudf/pylibcudf/strings/replace.pyx
index 5603ac849f0c..feb0e1c1d195 100644
--- a/python/pylibcudf/pylibcudf/strings/replace.pyx
+++ b/python/pylibcudf/pylibcudf/strings/replace.pyx
@@ -18,6 +18,10 @@ from pylibcudf.libcudf.strings.replace cimport (
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -29,7 +33,7 @@ cpdef Column replace(
     Scalar target,
     Scalar repl,
     size_type maxrepl=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Replaces target string within each string with the specified replacement string.
@@ -84,7 +88,7 @@ cpdef Column replace_multiple(
     Column target,
     Column repl,
     size_type maxrepl=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Replaces target string within each string with the specified replacement string.
@@ -137,7 +141,7 @@ cpdef Column replace_slice(
     Scalar repl=None,
     size_type start=0,
     size_type stop=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Replaces each string in the column with the provided repl string
diff --git a/python/pylibcudf/pylibcudf/strings/replace_re.pyx b/python/pylibcudf/pylibcudf/strings/replace_re.pyx
index 7112379911d5..05f2cb2153ab 100644
--- a/python/pylibcudf/pylibcudf/strings/replace_re.pyx
+++ b/python/pylibcudf/pylibcudf/strings/replace_re.pyx
@@ -16,6 +16,10 @@ from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.strings.regex_program cimport RegexProgram
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -27,7 +31,7 @@ cpdef Column replace_re(
     RegexProgram pattern,
     Scalar replacement=None,
     size_type max_replace_count=-1,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -83,7 +87,7 @@ cpdef Column replace_with_backrefs(
     Column input,
     RegexProgram prog,
     str replacement,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/reverse.pyx b/python/pylibcudf/pylibcudf/strings/reverse.pyx
index ba31d98f20b4..16d2dec30897 100644
--- a/python/pylibcudf/pylibcudf/strings/reverse.pyx
+++ b/python/pylibcudf/pylibcudf/strings/reverse.pyx
@@ -8,13 +8,17 @@ from pylibcudf.libcudf.column.column cimport column
 from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport reverse as cpp_reverse
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["reverse"]
 
-cpdef Column reverse(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column reverse(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Reverses the characters within each string.
 
     Any null string entries return corresponding null output column entries.
diff --git a/python/pylibcudf/pylibcudf/strings/slice.pyx b/python/pylibcudf/pylibcudf/strings/slice.pyx
index dc86db07ce93..de72105e0c03 100644
--- a/python/pylibcudf/pylibcudf/strings/slice.pyx
+++ b/python/pylibcudf/pylibcudf/strings/slice.pyx
@@ -19,6 +19,10 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
 from ..utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["slice_strings"]
@@ -28,7 +32,7 @@ cpdef Column slice_strings(
     ColumnOrScalar start=None,
     ColumnOrScalar stop=None,
     Scalar step=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Perform a slice operation on a strings column.
diff --git a/python/pylibcudf/pylibcudf/strings/split/partition.pyx b/python/pylibcudf/pylibcudf/strings/split/partition.pyx
index ea8735da729e..3e58c326e95a 100644
--- a/python/pylibcudf/pylibcudf/strings/split/partition.pyx
+++ b/python/pylibcudf/pylibcudf/strings/split/partition.pyx
@@ -13,6 +13,10 @@ from pylibcudf.libcudf.table.table cimport table
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.table cimport Table
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -24,7 +28,7 @@ __all__ = ["partition", "rpartition"]
 cpdef Table partition(
     Column input,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -75,7 +79,7 @@ cpdef Table partition(
 cpdef Table rpartition(
     Column input,
     Scalar delimiter=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/split/split.pyx b/python/pylibcudf/pylibcudf/strings/split/split.pyx
index 0a6c71ee77c6..3656e22af7e0 100644
--- a/python/pylibcudf/pylibcudf/strings/split/split.pyx
+++ b/python/pylibcudf/pylibcudf/strings/split/split.pyx
@@ -12,6 +12,10 @@ from pylibcudf.scalar cimport Scalar
 from pylibcudf.strings.regex_program cimport RegexProgram
 from pylibcudf.table cimport Table
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -33,7 +37,7 @@ cpdef Table split(
     Column strings_column,
     Scalar delimiter,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -87,7 +91,7 @@ cpdef Table rsplit(
     Column strings_column,
     Scalar delimiter,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -140,7 +144,7 @@ cpdef Column split_record(
     Column strings,
     Scalar delimiter,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -190,7 +194,7 @@ cpdef Column rsplit_record(
     Column strings,
     Scalar delimiter,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -241,7 +245,7 @@ cpdef Table split_re(
     Column input,
     RegexProgram prog,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -288,7 +292,7 @@ cpdef Table rsplit_re(
     Column input,
     RegexProgram prog,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -336,7 +340,7 @@ cpdef Column split_record_re(
     Column input,
     RegexProgram prog,
     size_type maxsplit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -380,7 +384,7 @@ cpdef Column split_record_re(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef Column rsplit_record_re(
-    Column input, RegexProgram prog, size_type maxsplit, object stream=None,
+    Column input, RegexProgram prog, size_type maxsplit, object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
@@ -425,7 +429,7 @@ cpdef Column rsplit_record_re(
 
 
 cpdef Column split_part(
-    Column input, Scalar delimiter, size_type index, object stream=None,
+    Column input, Scalar delimiter, size_type index, object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     cdef unique_ptr[column] c_result
diff --git a/python/pylibcudf/pylibcudf/strings/strip.pyx b/python/pylibcudf/pylibcudf/strings/strip.pyx
index 0e4dd400d53d..801b237a062e 100644
--- a/python/pylibcudf/pylibcudf/strings/strip.pyx
+++ b/python/pylibcudf/pylibcudf/strings/strip.pyx
@@ -15,6 +15,10 @@ from pylibcudf.libcudf.strings cimport strip as cpp_strip
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.strings.side_type cimport side_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -25,7 +29,7 @@ cpdef Column strip(
     Column input,
     side_type side=side_type.BOTH,
     Scalar to_strip=None,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Removes the specified characters from the beginning
diff --git a/python/pylibcudf/pylibcudf/strings/translate.pyx b/python/pylibcudf/pylibcudf/strings/translate.pyx
index 4c48fdd72ca3..fd911e362145 100644
--- a/python/pylibcudf/pylibcudf/strings/translate.pyx
+++ b/python/pylibcudf/pylibcudf/strings/translate.pyx
@@ -12,6 +12,10 @@ from pylibcudf.libcudf.strings cimport translate as cpp_translate
 from pylibcudf.libcudf.types cimport char_utf8
 from pylibcudf.scalar cimport Scalar
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 
@@ -45,7 +49,10 @@ cdef vector[pair[char_utf8, char_utf8]] _table_to_c_table(dict table):
 
 
 cpdef Column translate(
-    Column input, dict chars_table, object stream=None, DeviceMemoryResource mr=None
+    Column input,
+    dict chars_table: dict[int | str, int | str],
+    object stream: CudaStreamLike | None = None,
+    DeviceMemoryResource mr=None,
 ):
     """
     Translates individual characters within each string.
@@ -87,10 +94,10 @@ cpdef Column translate(
 
 cpdef Column filter_characters(
     Column input,
-    dict characters_to_filter,
+    dict characters_to_filter: dict[int | str, int | str],
     filter_type keep_characters,
     Scalar replacement,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """
diff --git a/python/pylibcudf/pylibcudf/strings/wrap.pyx b/python/pylibcudf/pylibcudf/strings/wrap.pyx
index f6180c7f56dd..23b20d8ac82c 100644
--- a/python/pylibcudf/pylibcudf/strings/wrap.pyx
+++ b/python/pylibcudf/pylibcudf/strings/wrap.pyx
@@ -9,6 +9,10 @@ from pylibcudf.libcudf.column.column_view cimport column_view
 from pylibcudf.libcudf.strings cimport wrap as cpp_wrap
 from pylibcudf.libcudf.types cimport size_type
 from pylibcudf.utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from rmm.pylibrmm.stream cimport Stream
 from cuda.bindings.cyruntime cimport cudaStream_t
@@ -16,7 +20,7 @@ from cuda.bindings.cyruntime cimport cudaStream_t
 __all__ = ["wrap"]
 
 cpdef Column wrap(
-    Column input, size_type width, object stream=None, DeviceMemoryResource mr=None
+    Column input, size_type width, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Wraps strings onto multiple lines shorter than `width` by
diff --git a/python/pylibcudf/pylibcudf/table.pyx b/python/pylibcudf/pylibcudf/table.pyx
index 3bbf318b6834..7c6d5b06c65b 100644
--- a/python/pylibcudf/pylibcudf/table.pyx
+++ b/python/pylibcudf/pylibcudf/table.pyx
@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
+from collections.abc import Sequence
 from cython.operator cimport dereference
 
 from cpython.pycapsule cimport (
@@ -34,6 +35,10 @@ from pylibcudf.libcudf.types cimport size_type
 from .column cimport Column
 from .types cimport DataType
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from pylibcudf._interop_helpers cimport (
     _release_schema,
     _release_array,
@@ -76,7 +81,7 @@ cdef class Table:
     """
     __hash__ = None
 
-    def __init__(self, columns, num_rows=None):
+    def __init__(self, columns: Sequence[Column], num_rows=None):
         columns = tuple(columns)
         if not all(isinstance(c, Column) for c in columns):
             raise ValueError("All columns must be pylibcudf Column objects")
@@ -127,7 +132,7 @@ cdef class Table:
     def from_arrow(
         obj: ArrowLike,
         dtype: DataType | None = None,
-        object stream=None,
+        object stream: CudaStreamLike | None = None,
         DeviceMemoryResource mr=None
     ) -> Table:
         """
@@ -359,7 +364,7 @@ cdef class Table:
         """The shape of this table"""
         return (self.num_rows(), self.num_columns())
 
-    cpdef Table copy(self, object stream=None, DeviceMemoryResource mr=None):
+    cpdef Table copy(self, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
         """Create a deep copy of the table.
 
         Parameters
@@ -404,7 +409,7 @@ cdef class Table:
 
         return PyCapsule_New(<void*>raw_schema_ptr, "arrow_schema", _release_schema)
 
-    def _to_host_array(self, object stream):
+    def _to_host_array(self, object stream: CudaStreamLike):
         cdef ArrowArray* raw_host_array_ptr
         cdef Stream _stream = _get_stream(stream)
         cdef cudaStream_t _cs = _stream.view().value()
diff --git a/python/pylibcudf/pylibcudf/table_equality.pyx b/python/pylibcudf/pylibcudf/table_equality.pyx
index e57ae491b76c..975bb90115c0 100644
--- a/python/pylibcudf/pylibcudf/table_equality.pyx
+++ b/python/pylibcudf/pylibcudf/table_equality.pyx
@@ -10,6 +10,10 @@ from rmm.pylibrmm.stream cimport Stream
 
 from .table cimport Table
 from .utils cimport _get_stream
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["tables_equal"]
@@ -19,7 +23,7 @@ cpdef bool tables_equal(
     Table left,
     Table right,
     null_equality nulls_equal=null_equality.EQUAL,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
 ):
     """Check if two tables are equal.
 
diff --git a/python/pylibcudf/pylibcudf/transform.pyx b/python/pylibcudf/pylibcudf/transform.pyx
index 66c392f524cd..2d29ecd4eef8 100644
--- a/python/pylibcudf/pylibcudf/transform.pyx
+++ b/python/pylibcudf/pylibcudf/transform.pyx
@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
+from collections.abc import Sequence
 from cython.operator cimport dereference
 
 from libcpp.memory cimport unique_ptr
@@ -31,6 +32,10 @@ from .expressions cimport Expression
 from .gpumemoryview cimport gpumemoryview
 from .types cimport DataType, null_aware, output_nullability
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 ctypedef const cpp_transform.transform_input const_transform_input
@@ -50,7 +55,7 @@ __all__ = [
 
 cpdef tuple[gpumemoryview, int] nans_to_nulls(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a null mask preserving existing nulls and converting nans to null.
@@ -92,7 +97,7 @@ cpdef tuple[gpumemoryview, int] nans_to_nulls(
 
 cpdef Column column_nans_to_nulls(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a column with nans converted to nulls.
@@ -129,7 +134,7 @@ cpdef Column column_nans_to_nulls(
 
 
 cpdef Column compute_column(
-    Table input, Expression expr, object stream=None, DeviceMemoryResource mr=None
+    Table input, Expression expr, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Create a column by evaluating an expression on a table.
 
@@ -166,7 +171,7 @@ cpdef Column compute_column(
 
 
 cpdef Column compute_column_jit(
-    Table input, Expression expr, object stream=None, DeviceMemoryResource mr=None
+    Table input, Expression expr, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """
     Create a column by evaluating an expression on a table
@@ -206,7 +211,7 @@ cpdef Column compute_column_jit(
 
 cpdef tuple[gpumemoryview, int] bools_to_mask(
     Column input,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a bitmask from a column of boolean elements
@@ -249,7 +254,7 @@ cpdef Column mask_to_bools(
     Py_ssize_t bitmask,
     int begin_bit,
     int end_bit,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Creates a boolean column from given bitmask.
@@ -292,13 +297,13 @@ cpdef Column mask_to_bools(
 
 
 cpdef Column transform(
-    inputs,
+    inputs: Sequence[Column],
     str transform_udf,
     DataType output_type,
     bool is_ptx,
     null_aware is_null_aware,
     output_nullability null_policy,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Create a new column by applying a transform function against
@@ -400,7 +405,7 @@ cpdef Column transform(
     return Column.from_libcudf(move(c_result), _stream, mr)
 
 cpdef tuple[Table, Column] encode(
-    Table input, object stream=None, DeviceMemoryResource mr=None
+    Table input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Encode the rows of the given table as integers.
 
@@ -437,7 +442,7 @@ cpdef tuple[Table, Column] encode(
 cpdef Table one_hot_encode(
     Column input,
     Column categories,
-    object stream=None,
+    object stream: CudaStreamLike | None = None,
     DeviceMemoryResource mr=None,
 ):
     """Encodes `input` by generating a new column
diff --git a/python/pylibcudf/pylibcudf/transpose.pyx b/python/pylibcudf/pylibcudf/transpose.pyx
index 38ad6e68e17f..eda2a5bc0f03 100644
--- a/python/pylibcudf/pylibcudf/transpose.pyx
+++ b/python/pylibcudf/pylibcudf/transpose.pyx
@@ -13,12 +13,16 @@ from rmm.pylibrmm.memory_resource cimport DeviceMemoryResource
 from .column cimport Column
 from .table cimport Table
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = ["transpose"]
 
 cpdef Table transpose(
-    Table input_table, object stream=None, DeviceMemoryResource mr=None
+    Table input_table, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Transpose a Table.
 
diff --git a/python/pylibcudf/pylibcudf/types.pyx b/python/pylibcudf/pylibcudf/types.pyx
index a5b6a19a0494..cbb369938a0b 100644
--- a/python/pylibcudf/pylibcudf/types.pyx
+++ b/python/pylibcudf/pylibcudf/types.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from libc.stddef cimport size_t
@@ -26,13 +26,10 @@ from pylibcudf.libcudf.types import order as Order  # no-cython-lint, isort:skip
 from pylibcudf.libcudf.types import sorted as Sorted  # no-cython-lint, isort:skip
 
 from functools import cache
+from typing import TYPE_CHECKING
 
-try:
+if TYPE_CHECKING:
     import pyarrow as pa
-    pa_err = None
-except ImportError as e:
-    pa = None
-    pa_err = e
 
 try:
     import pyarrow as pa
@@ -197,7 +194,7 @@ cdef class DataType:
         ret.c_obj = dt
         return ret
 
-    def to_arrow(self, **kwargs):
+    def to_arrow(self, **kwargs) -> pa.DataType:
         """
         Convert a datatype to arrow.
 
@@ -252,7 +249,7 @@ cdef class DataType:
                 )
 
     @staticmethod
-    def from_arrow(pa_typ) -> DataType:
+    def from_arrow(pa_typ: pa.DataType) -> DataType:
         """
         Construct a DataType from a Python type.
 
diff --git a/python/pylibcudf/pylibcudf/typing.py b/python/pylibcudf/pylibcudf/typing.py
new file mode 100644
index 000000000000..965bcd054054
--- /dev/null
+++ b/python/pylibcudf/pylibcudf/typing.py
@@ -0,0 +1,39 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+from typing import Protocol, TypeAlias, TypedDict
+
+from rmm.pylibrmm.stream import Stream
+
+
+class HasCudaStream(Protocol):
+    def __cuda_stream__(self) -> object: ...
+
+
+CudaStreamLike: TypeAlias = Stream | HasCudaStream
+
+
+class ArrayInterfaceBase(TypedDict):
+    data: tuple[int, bool]
+    shape: tuple[int, ...]
+    typestr: str
+    version: int
+
+
+class SupportsCudaArrayInterface(Protocol):
+    @property
+    def __cuda_array_interface__(self) -> ArrayInterfaceBase: ...
+
+
+class SupportsArrayInterface(Protocol):
+    @property
+    def __array_interface__(self) -> ArrayInterfaceBase: ...
+
+
+__all__ = [
+    "ArrayInterfaceBase",
+    "CudaStreamLike",
+    "HasCudaStream",
+    "SupportsArrayInterface",
+    "SupportsCudaArrayInterface",
+]
diff --git a/python/pylibcudf/pylibcudf/unary.pyx b/python/pylibcudf/pylibcudf/unary.pyx
index 37fda4c86291..3a14404c858b 100644
--- a/python/pylibcudf/pylibcudf/unary.pyx
+++ b/python/pylibcudf/pylibcudf/unary.pyx
@@ -18,6 +18,10 @@ from pylibcudf.libcudf.unary import \
 from .column cimport Column
 from .types cimport DataType
 from .utils cimport _get_stream, _get_memory_resource
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 from cuda.bindings.cyruntime cimport cudaStream_t
 
 __all__ = [
@@ -33,7 +37,7 @@ __all__ = [
 ]
 
 cpdef Column unary_operation(
-    Column input, unary_operator op, object stream=None, DeviceMemoryResource mr=None
+    Column input, unary_operator op, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Perform a unary operation on a column.
 
@@ -70,7 +74,7 @@ cpdef Column unary_operation(
     return Column.from_libcudf(move(result), _stream, mr)
 
 
-cpdef Column is_null(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_null(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Check whether elements of a column are null.
 
     For details, see :cpp:func:`is_null`.
@@ -102,7 +106,7 @@ cpdef Column is_null(Column input, object stream=None, DeviceMemoryResource mr=N
     return Column.from_libcudf(move(result), _stream, mr)
 
 
-cpdef Column is_valid(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_valid(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Check whether elements of a column are valid.
 
     For details, see :cpp:func:`is_valid`.
@@ -135,7 +139,7 @@ cpdef Column is_valid(Column input, object stream=None, DeviceMemoryResource mr=
 
 
 cpdef Column cast(
-    Column input, DataType data_type, object stream=None, DeviceMemoryResource mr=None
+    Column input, DataType data_type, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Cast a column to a different data type.
 
@@ -173,7 +177,7 @@ cpdef Column cast(
 
 
 cpdef Column bit_cast(
-    Column input, DataType data_type, object stream=None, DeviceMemoryResource mr=None
+    Column input, DataType data_type, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None
 ):
     """Bit-cast a column to a different data type.
 
@@ -210,7 +214,7 @@ cpdef Column bit_cast(
     return Column.from_libcudf(move(result), _stream, mr)
 
 
-cpdef Column is_nan(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_nan(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Check whether elements of a column are nan.
 
     For details, see :cpp:func:`is_nan`.
@@ -242,7 +246,7 @@ cpdef Column is_nan(Column input, object stream=None, DeviceMemoryResource mr=No
     return Column.from_libcudf(move(result), _stream, mr)
 
 
-cpdef Column is_not_nan(Column input, object stream=None, DeviceMemoryResource mr=None):
+cpdef Column is_not_nan(Column input, object stream: CudaStreamLike | None = None, DeviceMemoryResource mr=None):
     """Check whether elements of a column are not nan.
 
     For details, see :cpp:func:`is_not_nan`.
diff --git a/python/pylibcudf/pylibcudf/utils.pyx b/python/pylibcudf/pylibcudf/utils.pyx
index 314e62f77606..576351dc798e 100644
--- a/python/pylibcudf/pylibcudf/utils.pyx
+++ b/python/pylibcudf/pylibcudf/utils.pyx
@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 from cython.operator import dereference
@@ -16,10 +16,14 @@ from rmm.pylibrmm.memory_resource cimport (
     get_current_device_resource,
 )
 
-from rmm.pylibrmm.stream import DEFAULT_STREAM, PER_THREAD_DEFAULT_STREAM
+from rmm.pylibrmm.stream import DEFAULT_STREAM, PER_THREAD_DEFAULT_STREAM, Stream
 
 
 import os
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pylibcudf.typing import CudaStreamLike
 
 
 # Check the environment for the variable CUDF_PER_THREAD_STREAM. If it is set,
@@ -47,7 +51,7 @@ cdef vector[reference_wrapper[const scalar]] _as_vector(list source):
     return c_scalars
 
 
-cpdef Stream _get_stream(object stream = None):
+cpdef Stream _get_stream(object stream: CudaStreamLike | None = None):
     if stream is None:
         return CUDF_DEFAULT_STREAM
     if isinstance(stream, Stream):