Harun

I’m an Information Engineering undergrad at Universitas Gadjah Mada, and I’ve spent the last few years shipping production web applications — mostly Next.js and TypeScript, usually under a deadline, occasionally for ten thousand people at once.

Lately I’ve been pulled toward the other side of the problem: how agentic systems evaluate themselves. I write and benchmark frontier-level coding tasks for AI models, which turns out to be a very effective way of learning where software engineering judgement actually lives.

I also like the unglamorous half of the job — typed API contracts, CI/CD that doesn’t lie to you, and getting a box to serve reliably at 3am with no network.

See my work   Read my resume