Hello and welcome! I am currently a Resident at Anthropic, where I perform research in how to build powerful AI systems that are safe, reliable, and interpretable.

My current research interests are broadly in understanding the capabilities of large language models, and generally how to align models such that they are both helpful and safe. I also have a parallel interest in machine translation and multilingual NLP, with a focus on making the benefits of AI more broadly accessible.

My resume is available here (last updated December 22, 2022). I'll be updating this very regularly over the course of the next year.

Most recently, I was a PhD student at UC Berkeley in the Physics Department advised by Raphael Bousso. My research focused on understanding the interplay between quantum information and quantum gravity. To that end, my work looked at various aspects of holography (AdS/CFT), quantum error correcting codes, entanglement measures, and random tensor networks.

I did my undergraduate studies at Stanford University, where I was fortunate enough to work with and write an honors thesis under Shamit Kachru, studying string-theoretic aspects of black holes.

I hold a long-standing interest in physics and mathematics education, and am passionate about using research-driven pedagogy tools to more effectively teach these topics.

If you'd like to chat about any of the things above (or most other topics), feel free to drop me a line at my email: newtoncheng(at)berkeley.edu.