About

I’m Pilkwang Kim. I earned a Ph.D. in theoretical physics from Seoul National University (SNU), where I worked on density functional theory and topological materials, and I now work on semiconductor device simulation at Samsung.

I write here about three things, mostly: physics, where it meets machine learning, and whatever I’ve been thinking through lately.

What I work on

I work on semiconductor devices — mostly the why behind how they behave. Day to day that means device physics, characterization, and the kind of judgment that helps a development effort understand what’s actually happening and decide what to do about it. Simulation, TCAD among other tools, is one way I get there; just as often it comes down to knowledge and reasoning — understanding the physics well enough to see the answer before a solver would. That reasoning is the part I care about most, and where the physics background pulls its weight.

Machine learning

I think many ML problems are physics problems in disguise — that the right structural prior often matters more than model complexity. I keep testing that idea across domains, building baselines whose job is to make the underlying structure visible rather than to chase a score. The same instinct carries into my work: a surrogate model is only as trustworthy as its worst-behaved corner of input space, and physics is often the cheapest way to find those corners.

Beyond work

Away from the screen I’m an ultramarathoner and trail runner, with swimming and weightlifting filling in the rest. I’m happiest running long distances on my own — it’s the closest thing I have to a meditation, and where a lot of my thinking quietly sorts itself out.

Get in touch

I’m always glad to talk about physics-informed modeling, feature engineering, or surrogate models for scientific simulation.