How kartikeya-0211/quantum-tutor works
Project type and purpose
• This is a full-stack RAG (Retrieval-Augmented Generation) chatbot, “Quantum Tutor,” built in Python.
• It provides an interactive Streamlit UI (app.py) that connects to a LangChain-based RAG pipeline, uses ChromaDB for embeddings, and dispatches LLM requests via LiteLLM to external providers (Groq, Ollama).
File structure
• Top-level:
• app.py – Streamlit entrypoint for the web UI
• chains/ – RAG pipeline definitions (chains/rag_chains.py)
• loaders/ – Document ingestion (loaders/load_documents.py)
• retrievers/ – Re-ranking logic (retrievers/reranker.py)
• utils/ – Helpers (chunker, embedder, llm_loader, query_classifier)
• config/ – litellm_config.yaml for LLM provider settings
• data/ – PDF knowledge base
• tests/ – Benchmark harness and evaluation scripts
• requirements.txt, .env.example – dependencies and env setup
• Architectural patterns:
• A clear separation between “frontend” (Streamlit UI) and “backend” (RAG pipeline).
• A single monolithic service with modular subcomponents rather than microservices.
README for additional insights
• Core technologies: Streamlit, LangChain, LiteLLM, ChromaDB.
• Use cases: querying academic materials, benchmarking LLMs.
• Frontend (blue),
• Backend pipeline (green),
• Data store (orange),
• External services (purple).
• Rectangles for code modules,
– Increase font size on labels for readability.