AI RAG Starter

Production-ready Retrieval-Augmented Generation (RAG) API built with FastAPI, ChromaDB, and OpenAI-compatible LLMs.

Features

• Document ingestion — PDF, TXT, Markdown via LangChain loaders
• Vector store — ChromaDB with persistent storage
• Embeddings — OpenAI or any OpenAI-compatible endpoint (Ollama, vLLM)
• RAG pipeline — Chunk → embed → retrieve → generate with citations
• REST API — /ingest, /query, /health endpoints
• Docker — One-command local deployment

Architecture

Documents → Chunker → Embeddings → ChromaDB
                                      ↓
User Query → Embed → Retrieve top-k → LLM → Answer + Sources

Quick Start

cp .env.example .env   # add OPENAI_API_KEY
pip install -r requirements.txt
uvicorn app.main:app --reload

# Ingest documents
curl -X POST http://localhost:8000/ingest \
  -H "Content-Type: application/json" \
  -d '{"text": "Kubernetes is a container orchestration platform.", "source": "docs"}'

# Query
curl -X POST http://localhost:8000/query \
  -H "Content-Type: application/json" \
  -d '{"question": "What is Kubernetes?"}'

Docker

docker compose up --build

API

Endpoint	Method	Description
/health	GET	Health check
/ingest	POST	Add text to vector store
/ingest/file	POST	Upload and ingest a file
/query	POST	RAG question answering

Configuration

Variable	Default	Description
OPENAI_API_KEY	—	API key for embeddings + LLM
OPENAI_BASE_URL	https://api.openai.com/v1	Compatible API base URL
EMBEDDING_MODEL	text-embedding-3-small	Embedding model
LLM_MODEL	gpt-4o-mini	Chat model
CHROMA_PERSIST_DIR	./chroma_data	Vector store path

License

MIT © almightymoon