← All projects
RAG Document Intelligence
A retrieval-augmented search tool over publishing content, serving 200 QPS in production.
- OpenAI
- ChromaDB
- Node.js
- AWS
Built at Compro Technologies. The code is proprietary.
A document intelligence tool for the CUP platform that combines an OpenAI LLM with ChromaDB vector search to answer questions over large document sets.
Highlights
- RAG pipeline: OpenAI LLM + ChromaDB vector store.
- Supports 200 QPS in production.
- Improved search accuracy across the CUP platform.