Portfolio RAG
Production-ready retrieval-augmented generation with automated evaluation
Reference architecture for reliable RAG featuring end-to-end ingestion, chunking, embedding, and retrieval using LangChain LCEL. Self-hosted Qdrant vector database in Docker. Comprehensive evaluation harness proving Recall@4 = 100%, citation-backed answers to prevent hallucinations, and GitHub Actions CI/CD with automated integration tests. Demonstrates clean architecture, source grounding, and reproducible infrastructure.
Tech Stack
Python
LangChain
Qdrant
Streamlit
Docker
OpenAI
pytest