I'm Anudeep, an AI engineer and product builder in Austin. I build retrieval systems, agent tools, and the applications around them. I like work you can inspect: a source passage, a tool call, an evaluation, a working interface.
Portfolio: ten projects and their demos · LinkedIn · Writing
Document retrieval and saved conversations through MCP and FastAPI. Start with the provider-free walkthrough, or try the browser citation demo. The demo matches words in synthetic notes and shows the source line.
Research code for evaluating retrieval, answer quality, and latency in smoking-cessation question answering. The synthetic example runs separately from the study data.
A campaign-planning prototype: brief, sample scripts, review, and a simulated render queue. Try the workflow. It doesn't generate videos.
Multi-engine web extraction with cleaned content and streaming progress. The repository covers the API, browser adapters, and setup.
Delivery planning, capacity calculations, and saved drafts. The standalone capacity demo shows its formulas and saves locally in your browser; the full application uses React, Express, and SQLite.
Messaging protocols for research teams: templates, language variants, and conditional sequences. Explore the public illustration. Live study delivery is separate.
On the wider bill: Polaris, Gimmie, Field Notes, Vedam, exec.ai, and Purna. Each case study explains my contribution and what you can try today.
Python · TypeScript · C++ · React · FastAPI · MCP
For the C++ side: an LSTM built from the standard library.