Hi, I'm Tarun.

I'm a full-stack developer turned AI systems engineer, based in Hyderabad, India.

I spent four years building production mobile apps with Flutter — the kind that sit on real users' phones and have to work even when the network doesn't. Apps on the App Store. Apps with native bridges into HealthKit and Google Fit. Apps with real-time chat and background location.

Then I went deeper. I started building the AI side of products instead of just calling AI APIs. I shipped a production translation pipeline running Gemma 4 31B on a GPU server — 60,000 verses of the Bible translated with zero errors, using a custom DAG, GPU mutex gating, and a guardrail I had to invent for the problem.

These days I sit at a rare crossroads. Most engineers can build the app, or the AI behind it. I do both, on the same project, by myself. It means a startup or a client can hire one person instead of two — and the AI side actually talks fluently to the product side, because the same brain wrote both.

I write about what I learn — engineering case studies, architecture migrations, and reflections on what it's like to work with AI every day without losing the muscle of thinking for yourself.