I’m Oz — an AI safety researcher with a background in particle physics

I am currently a fellow at CBAI, working with Hadas Orgad.

Right now my research interests revolve around personas, subliminal learning, and mechanistic interpretability. You can read about some of my projects on my AI safety page.

In my former physics life, I was a Wilson Fellow at Fermilab, a tenure-track position equivalent to assistant professor, working on the CMS experiment at the Large Hadron Collider. I joined Fermilab as a postdoc after completing my PhD in physics at Johns Hopkins University in 2022, and started the fellowship in 2026.

I specialized in applying ML methods to analyze complex particle physics data, with the hope of answering deep questions about the fundamental nature of the universe. I was particularly excited about anomaly detection and using it to find unexpected new phenomena hiding in massive collider datasets. See my physics projects for more.

Check out my LessWrong profile for my recent AI safety work. I also have a blog on Substack.