Most RAG (Retrieval-Augmented Generation) projects you see online are great demos.

But try running them in production and you’ll quickly hit issues:

• no ingestion pipeline
• no async processing
• no scaling story
• no observability
• no proper deployment setup

So I decided to build something that actually works beyond demos.

Introducing Ragify

An open-source, production-oriented RAG backend built with:

• Node.js + Express + TypeScript
• MongoDB for documents + logs
• Qdrant for vector search
• Redis + BullMQ for async ingestion
• OpenAI for embeddings + responses

GitHub: https://github.com/open-loft/ragify

What makes it different

Instead of just “chat + embeddings”, Ragify focuses on the full pipeline:

Upload → Queue → Chunk → Embed → Store → Retrieve → Generate

Some key features:

• Async ingestion (doesn’t block uploads)
• Token-based chunking with overlap
• Streaming responses (SSE)
• Rate limiting + config validation
• Dockerized production setup

Why I built this

I wanted a backend that:

• I could self-host
• I could extend safely
• I could actually use in a real product

Looking for feedback / contributors

Would love input on:

• improving retrieval quality
• reranking approaches
• hybrid search strategies
• cost + latency optimization

If you’re building in the RAG / LLM space, this might be useful.

Would appreciate your thoughts 🙌