Tetsuto Nagashima

Hi, I’m Tets, a first-year M.S. Computer Science student at USC working toward research in AI safety, especially interpretability, evaluations, representation-level analysis, and adversarial robustness. I am drawn to problems where the goal is not only to make a model perform better, but to understand the structure behind its behavior, when it fails, and how to intervene.

Before starting my M.S. CS, I completed a B.S. in Physics/Computer Science, a B.A. in Philosophy and Physics, and a concurrent M.S. in Mechanical Engineering at USC. That nonlinear path still shapes how I work: physics and philosophy made me comfortable with opaque systems, while ML engineering taught me how to build, test, and diagnose them in practice.

I am currently a Student Researcher in USC’s DILL Lab under Swabha Swayamdipta, working on fine-grained evaluations of reasoning in large language models. Recent work includes data poisoning in performative prediction, subliminal learning experiments from an Apart Research hackathon, mechanistic-interpretability work trials around decompositions and superposition, lab-automation systems at Tetsuwan Scientific, and ML curriculum for PhD students and industry professionals through USC’s AI for Discovery in the Sciences workshop.

Longer term, I also want to connect technical safety work with public AI education. I think better non-technical understanding of what AI systems can and can’t do is important for responsible use, democratic governance, and public discourse that is grounded rather than hype- or fear-driven.

You can find more on my CV, my projects, and occasional writing on the blog. If any of this overlaps with what you are working on, I’d be happy to get in touch.