AI Chatbot — RAG · Agents · Tools
A production retrieval-augmented chatbot over internal documents. Metadata-filtered retrieval feeds a tool-using agent loop; FastAPI exposes the API; the Python service runs on Render.
- Year
- 2025
- Type
- Day job
- Role
- Solo developer · Production
- Status
- In production
What needed solving
Internal teams kept hitting the same long handbooks and policy docs for the same questions. Plain keyword search missed context, and single-shot RAG over the whole corpus returned irrelevant chunks when a question touched only one section.
The solution
Built a metadata-aware RAG pipeline: each chunk carries source / section / tag metadata used to pre-filter the candidate set before semantic search. On top, an agent loop (LangChain tools + structured tool-calling) decides when to retrieve, when to summarize, and when to call structured-extraction tools. Exposed via a FastAPI service deployed on Render with streaming responses.
What changed
- Metadata filtering narrows the candidate set before vector ranking — fewer off-topic chunks reach the LLM
- Agent with tools handles structured questions ("extract X from section Y") beyond plain semantic search
- Streaming responses + per-request tracing make latency and quality issues observable in production
Technical highlights
Need something like this?
I take on a small number of projects each quarter. Let's talk if your idea fits.
News Desk — Autonomous Story-Discovery Agent
An automated newsroom agent that reproduces an editorial team's morning routine — searches the web and ~52 curated sources across 11 countries for textile/apparel news, scores and de-dupes the best stories with an LLM, confirms each against its primary source, then drafts publish-ready articles. Two-phase with a human in the loop, streamed live over SSE.