👋 Hi, I'm Jerry.

I'm an AI safety researcher at Anthropic. I lead the deployment-robustness team, building production safety systems to defend Claude from being misused. I'm based in San Francisco.

I was previously a research engineer at Google DeepMind and a student researcher at Google Brain. I also studied at Stanford University as an undergraduate specializing in AI. In 2022, I interned at Meta as a software-engineering intern.

Socials

Selected papers (all papers).

Constitutional Classifiers++: Efficient Production-Grade Defenses against Universal Jailbreaks. blog
Constitutional classifiers: Defending against universal jailbreaks across thousands of hours of red teaming. blog
Long-form factuality in large language models. code
Simple synthetic data reduces sycophancy in large language models. code
Symbol tuning improves in-context learning in language models. blog
Larger language models do in-context learning differently. blog

Talks.

June 2026The Real World AI Security Conference.
May 2026Caltech AI Alignment.
March 2026Princeton AI Tiger Trek.
June 2025Caltech AI Alignment.
April 2025Generative AI Summit - Silicon Valley.
March 2025Princeton AI Alignment. youtube
March 2025Princeton AI Tiger Trek.
Sept. 2023Google CS Research Mentorship Program.

Last updated September 2026.