A server-side CKKS GPU library fully interoperable with OpenFHE.
-
Updated
Nov 2, 2026 - Cuda
A server-side CKKS GPU library fully interoperable with OpenFHE.
Open-source FHE client and toolchain. Build fully homomorphic encryption applications with the nb DSL, instrumented OpenFHE, FHETCH API or a CUDA-style library API, record one Polynomial IR trace, and deploy to the Niobium accelerator
Harness and example implementation the FHE fetch-by-similarity workload
Build apps that compute on data no one can see, not even the server running them. Agent skills that guide your AI coding agents from privacy idea to a working encrypted application.
Mamba-2-130M encrypted inference research with CKKS on DGX Spark. Python reference, native GPU kernel, and reproducible experiments.
Reference implementation of the BHDR regression kernel: BSGS-hoisted diagonal Kernel SHAP regression under CKKS FHE.
Drop-in encrypted Fairlearn metrics over CKKS. Same API surface; ciphertext arithmetic via TenSEAL or OpenFHE.
Unified attack-replay regression harness for FHE libraries (SEAL, OpenFHE, Lattigo, tfhe-rs).
FHE Oracle — adversarial precision testing for Fully Homomorphic Encryption. Open-source edition (AGPL-3.0).
Confidential computation and governance for AI across organizations, keep sensitive data and models private, enforce purpose-bound release, and produce auditable evidence.
Run a neural-network intrusion detector on encrypted network traffic with FHE, executed on the Niobium Fog. Apache-2.0 sample code + public SDK; Fog execution is a hosted service and needs a Niobium account.
Starter kit for encrypted (FHE) computation on the Niobium Fog. Apache-2.0 sample code + public SDK; Fog execution is a hosted service and needs a Niobium account.
Score a card transaction for fraud without the scoring party ever seeing it — a neural network under FHE, running on a Niobium FPGA over the Fog.
FPGA accelerator for exact OpenFHE DCRTPoly polynomial multiplication in Z_q[X]/(X^4096+1), on a PYNQ-Z2 (XC7Z020)
30-day hospital readmission risk scored under fully homomorphic encryption. The patient is analyzed in full. The model reads nothing at all.
A Benchmarking Framework for Fully Homomorphic Encryption Libraries
SafeAdapt-VANET: contract-constrained hierarchical LLM agent orchestration for privacy-preserving federated learning in dynamic vehicular networks (research prototype)
Docker base image with pre-built OpenFHE libraries for creating language bindings and applications. Ready-to-use development environment for homomorphic encryption projects.
To associate your repository with the openfhe topic, visit your repo's landing page and select "manage topics."