Applied AI/ML experiments by @evanallen13 — retrieval, agents, and the Model Context Protocol.
- Retrieval & RAG — hybrid vector + BM25 over
pgvector, local embeddings via Ollama - AI agents — tool-using LLM loops with tracing and observability
- MCP servers — exposing tools and search to Claude, Cursor, and other MCP clients
- AI infrastructure — Kubernetes, GitHub Actions, and self-hosted runners for AI workloads
- todo-mcp — MCP server with OpenTelemetry tracing
- BM-25-Search (sister org) — MCP server + CLI for BM25 search over a Markdown corpus
- Rag-Chat-Bot (sister org) — GPT-4 + ChromaDB RAG over CSV chat data
Earlier work lives at MandalAutomations — RAG chatbots, embedding visualizers, and GitHub-native AI tooling.