How brocoder07/docuchat---rag-application works
Project type and purpose
• This is a full-stack Retrieval-Augmented Generation (RAG) application named DocuChat.
• Purpose: Let users upload multi-format documents, process and store them as embeddings, then query them conversationally via a local LLM.
File structure
• Top-level directories:
• src/api → FastAPI backend (routes, models, services)
• src/frontend → Streamlit UI (components, services, utils)
• src/context, src/document_processor, src/embedding_manager, src/chroma_manager, src/llm_integration, src/rag_pipeline → Core RAG pipeline modules
• data/vector_store → ChromaDB persistence
• scripts → Utility/testing scripts
• config files (.streamlit, src/config.py, src/api/core/config.py)
• Architectural patterns:
• Layered separation: frontend ↔ backend ↔ pipeline ↔ persistence ↔ external LLM
• Modular pipeline stages for chunking, embedding, storage, retrieval, and LLM inference.
README insights
• Architecture overview confirms 4 layers:
1. Streamlit UI
2. FastAPI API
3. RAG Pipeline (Document Processor, Embedding Manager, ChromaDB, Ollama LLM)
4. Data store & external service (ChromaDB, Ollama)