Case Study

DocuMind AI — Document Q&A SaaS (RAG)

Multi-tenant document-Q&A SaaS built on a production RAG pipeline. Upload PDFs/text → chunked, embedded, and stored in PostgreSQL with pgvector → ask questions in natural language → get answers grounded in your data with inline source citations. Engineered like a real product: JWT auth with per-tenant isolation, async ingestion via Celery/Redis, per-user rate limiting, Stripe subscription billing, streaming responses, and Langfuse LLM observability. FastAPI + SQLAlchemy async + LangChain, with a React frontend, Dockerized stack, pytest suite, and GitHub Actions CI.