dev-it-with-me/RagUltimateAdvisor

Implements a full-stack RAG pipeline using LlamaIndex for document processing, PostgreSQL with pgvector for semantic search, and Ollama for local LLM inference, eliminating cloud dependencies. The architecture chains PDF ingestion → vector embeddings → similarity-based retrieval → context-aware generation, with FastAPI backend and React frontend communicating via typed endpoints. Includes Docker Compose orchestration for easy deployment and supports swappable models (Llama 3.2, Mistral, Gemma) for customization.