Frank Nakasako

I studied pure math at Berkeley, and now I am funded by Coefficient Giving to work on scalable alignment. My primary focus is AI generalization and understanding emerging multi-actor environments.

Publications

• Encoded Coordination on the Open Web. Ethan Elasky, Can Küçükkurt, Frank Nakasako, David Africa. LessWrong, 2026.
• Debate Helps Weak Judges Reward Stronger Models. Ethan Elasky, Frank Nakasako, Naman Goyal. arXiv, 2026.

About me

I like literature. I like the KJV bible, William Faulkner, Paradise Lost, The Faerie Queene, and Virginia Woolf. I like working on interesting things. Right now that is trying to better understand intelligence—machine and biological.

email | github | x