A full-stack Retrieval-Augmented Generation system with:
- β Multi-workspace knowledge bases
- π PDF ingestion + citation viewer
- π URL ingestion
- π Semantic vector search (Qdrant)
- π§ Llama-3 via Ollama
- β‘ FastAPI backend
- π¨ Next.js frontend
- π³ Docker production stack
| Layer | Tech |
|---|---|
| Frontend | Next.js + Tailwind + Shadcn + React PDF |
| Backend | FastAPI + Sentence Transformers |
| Vector DB | Qdrant |
| LLM | Ollama (Llama-3) |
| Deployment | Docker Compose |
npm run dev # frontend
uvicorn api:app --reload # backend
docker compose up qdrant # dbπ³ Production
docker compose -f docker-compose.prod.yml up -d --build