Production RAG with Hybrid Search

Combine dense vector search with BM25 keyword search. Add query rewriting, hypothetical document embedding (HyDE), streaming response, and a full eval suite.

You will learn to:

• Combine dense vector search and BM25 keyword search using Reciprocal Rank Fusion
• Implement query rewriting to improve retrieval quality for ambiguous queries
• Apply HyDE (Hypothetical Document Embedding) to boost semantic search precision
• Build a streaming FastAPI + Streamlit layer on top of a production RAG system
• Build a streaming FastAPI + Streamlit layer on top of a production RAG system