A complete Retrieval-Augmented Generation (RAG) setup running locally with Docker Compose.

• Ollama - Local embeddings using mxbai-embed-large
• PostgreSQL + pgvector - Vector database for semantic storage
• Flask Web App - Simple interface to store and search documents
• Docker Compose - Orchestrates all services

• Docker & Docker Compose installed
• At least 4GB free disk space (for models and data)

docker compose up --build

Open your browser to: http://localhost:8000

docker exec -it rag_ollama ollama pull mxbai-embed-large

• 8000: Web application
• 11434: Ollama API
• 5432: PostgreSQL database

CREATE TABLE documents (
    id SERIAL PRIMARY KEY,
    text TEXT,
    embedding VECTOR(1024)  -- mxbai-embed-large produces 1024-dim vectors
);