Hi, I’m Jihun Yoon, an AI Research Engineer based in Seoul, South Korea.

This is my technical notebook: a place to think through and document AI research, systems, experiments, and the work in between.

My current work focuses on agentic systems—particularly harness engineering, self-improving agents, and AI scientist systems. Previously, I worked on video perception for computer-assisted surgery, with a focus on data-efficient learning and world models.

Across these areas, I’m interested in how AI systems learn useful representations, reason about complex environments, and turn that understanding into reliable action. I write about research questions, projects, implementation details, and lessons learned while building.

Outside of work, I enjoy studying and building quantitative trading systems. I live in Seoul with my wife and our puppy, Rungji. I used to go bouldering four days a week; these days, I mostly take walks with Rungji instead.