Hi @bharagha @14pankaj @jgespino, I have a proposal for following enhancement, please assign this issue to me

1. Description of the Enhancement

This enhancement proposes upgrading the chat-question-and-answer pipeline from a dense-only retrieval (PGVector MMR) to a Hybrid Retrieval strategy. By integrating a sparse BM25 retriever with the existing vector search, the application will achieve significantly higher accuracy for keyword-specific queries.

Technical Overview:

• Integrate LangChain's EnsembleRetriever to fuse Dense and Sparse search results.
• Add rank-bm25 as a dependency in pyproject.toml.
• Configure weighted scoring via environment variables to balance semantic vs. lexical relevance.

2. Use Cases and Benefits

• Technical Accuracy: BM25 provides precise matches for specific keywords, product codes, and rare technical terms (e.g., "OpenVINO-2026") where dense embeddings might be imprecise.
• Robust Retrieval: Provides a "safety net" for short queries or terms not well-represented in the embedding model's vocabulary.
• Industry Standard: Demonstrates RAG best practices using state-of-the-art hybrid search techniques.

3. Additional Context or References

• LangChain Ensemble Retriever: Documentation
• Target Component: app/chain.py
• This upgrade ensures the Intel Edge GenerativeAI Suite's sample applications remain benchmark-competitive and robust for technical documentation use cases.