Full-stack RAG system for querying research papers via natural language.
- FastAPI backend with LangGraph pipeline (route → retrieve → grade → generate → cite)
- 4-step ingestion pipeline with MongoDB checkpointing (parse, clean, chunk, index)
- ChromaDB vector store with BAAI/bge-base-en-v1.5 local embeddings
- Groq LLM (llama-3.3-70b-versatile) for routing, grading, and generation
- Paper ingestion via PDF upload or arXiv ID fetch
- Streamlit frontend with SSE streaming chat and papers library sidebar
- Docker Compose setup for full stack deployment
- pytest suite: unit, integration, and smoke tests
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
