Skip to content

feat(rag): FAISS vectors and hybrid retrieval #12

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 18 commits into
base: main
Choose a base branch
from
Open

feat(rag): FAISS vectors and hybrid retrieval #12

wants to merge 18 commits into from

Conversation

@haz3141
Copy link
Owner

@haz3141 haz3141 commented Sep 6, 2025

Step 6: Advanced RAG Enhancements

Features Added

  • Embeddings Pipeline: Added Embeddings class with sentence-transformers support and stub fallback for offline testing
  • FAISS Vector Store: Added FaissIndex for efficient vector similarity search with numpy fallback
  • Hybrid Retriever: Combined BM25 and vector scores with configurable alpha weighting
  • MCP Endpoint: Added /tools/retrieve_hybrid endpoint for hybrid retrieval
  • Comprehensive Tests: Added tests for hybrid retrieval with deterministic stub embeddings
  • CI Updates: Updated CI to run hybrid retrieval tests with stub embeddings

Technical Details

  • Uses sentence-transformers for embeddings (with fallback to stub for tests)
  • FAISS for vector similarity search (with numpy fallback)
  • Hybrid scoring combines normalized BM25 and vector scores
  • All tests use stub embeddings for deterministic, offline testing
  • Environment variables for configuration (EMBED_BACKEND, EMBED_MODEL, HYBRID_ALPHA)

Version

Bumps to v0.5.0

Testing

  • All tests pass with stub embeddings
  • Hybrid endpoint tested and working
  • CI configuration updated to include hybrid tests

haz3141 added 18 commits September 6, 2025 16:20
@haz3141
chore(deps): add dspy-ai and pin versions
b0ef6d5
@haz3141
feat(dsp): add Summarize module and demo runner
2ce1db9
@haz3141
feat(mcp): expose /tools/summarize backed by DSPy module
10c3a53
@haz3141
test(dsp): add unit test and HTTP smoke test
6e0a6fe
@haz3141
chore(env): add DSPY_MODEL placeholder to .env.sample
0d6baee
@haz3141
chore(version): bump to 0.3.0 for DSPy integration
538daeb
@haz3141
ci: add minimal lint+unit test workflow
0cc896c
@haz3141
feat(rag): scaffold ingestion, chunking, BM25 retriever with fixture
6c62483
@haz3141
test(rag): add deterministic unit test for chunking and BM25 retrieval
c5a60df
@haz3141
feat(mcp): add /tools/retrieve using minimal BM25 over lab fixtures
048d069
@haz3141
chore(env): add RAG_DATA_DIR and RAG_TOP_K placeholders
1613c8c
@haz3141
ci: run RAG unit test alongside summarize test
08a1556
@haz3141
chore(version): bump to 0.4.0 for minimal RAG endpoint
38b1a2f
@haz3141
feat(rag): add embeddings pipeline, FAISS vector store, and hybrid re…
e79b004 
…triever

- Add Embeddings class with sentence-transformers and stub fallback
- Add FaissIndex for vector similarity search with numpy fallback
- Add HybridRetriever combining BM25 and vector scores
- Add comprehensive tests for hybrid retrieval
- Update requirements.txt with new dependencies
@haz3141
feat(mcp): add /tools/retrieve_hybrid endpoint
92c6993 
- Add RetrieveHybridRequest model with query, k, and alpha parameters
- Add retrieve_hybrid endpoint that uses HybridRetriever
- Returns hits with both vector and BM25 scores for transparency
@haz3141
chore(env): add EMBED_BACKEND, EMBED_MODEL, HYBRID_ALPHA placeholders
4017dfd 
- Update .env.sample with new environment variables for embeddings and hybrid retrieval
- Update CI configuration to run hybrid retrieval tests with stub embeddings
@haz3141
chore(version): bump to 0.5.0 for advanced RAG (FAISS + hybrid)
043e22c
@haz3141
feat(security): add Guardian allowlist + PII redaction
3107431 
- Add lab/security/redact.py for PII pattern redaction
- Add lab/security/guardian.py for tool allowlist and response sanitization
- Add comprehensive tests for security modules
- Integrate Guardian into MCP server endpoints
- Add audit logging for all tool calls
- Add evaluation harness with hit@k and mrr@k metrics
- Add observability with JSONL audit logs
- Add promotion flow documentation and security checklist
- Update CI with new tests and evaluation step
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@haz3141