Engineering with a researcher's attention to detail.

I work across the boundary between understanding an ML system and building one that holds up in practice. My current interests include scientific machine learning, protein and biological foundation models, geometric representations, and the systems that serve modern models reliably.

I like to start close to the underlying assumptions: what a representation preserves, where an objective comes from, what the computational bottleneck is, and how a design changes once it leaves an experiment. I also care about the system around a model—how it is operated, modified, observed, and scaled responsibly.

Understand the mathematics. Understand the model. Understand the system.

For code, visit GitHub. For short notes and conversation, find me on X.