VA-chat / RAG
Staff-facing AI document analysis and chat for VA packets: upload and OCR, analysis, and advisor-framed chat over Azure AI Foundry plus a vector knowledgebase, with admin and mass-train surfaces on a Docker / nginx path.
Not the original build. The company VA-chat / RAG platform was brought to a working production path — ops, wiring, reliability, the live route. Not a from-scratch RAG design claim.
Role Production ownership of an inherited AI stack.
What this proves Inherited AI stack driven to a live path — Azure AI and Python service literacy, plus production judgment around RAG.
- Python
- FastAPI
- Chainlit
- Docker
- nginx
- Postgres
- Azure AI Foundry
- OpenAI-family models
- JWT
Public Internal staff tool. No project URL or store IDs on this page.
Full case study coming soon. This route is in place so the work can grow without rewriting the homepage.