Shunyu Yao

Senior Staff Research Scientist@Google DeepMind, previously(@Anthropic@UCBerkeley@Stanford@Tsinghua)

About Me

I'm currently a research scientist at Google DeepMind

I was a research scientist at Anthropic. My research at Anthropic is centered around large scale reinforcement learning of large language models, mostly focus on improving model capabilities.

Previously, I was a theoretical physicist. I did a short postdoc at Berkeley Center for Theoretical Physics before Anthropic. I did my PhD at Stanford Institute for Theoretical Physics with Douglas Stanford and Stephen Shenker, and my Bachelor in IAS, Tsinghua with Zhong Wang

Projects(@Anthropic)

My main contribution @anthropic includes:

Agentic coding/tool-use in large scale RL(these research comes into Claude 3.7)

RL numerics control and basic algorithm design(these research comes into Claude 4 family)

New generation of RL algorithm

Fundamental science on RL hyper parameter

Science of SL and its impact on RL

Projects(@Physics)

@Hep-th: My work focused on dynamics of complex quantum systems, including quantum chaos, quantum blackholes and their relation to quantum information. eg. the development of scramblon theory.

@CMT: My work focused on open quantum systems and topological phase of matter. eg. The discovery of non-Hermitian Skin effect

Resume

Download My Resume (PDF)

Latest Posts

Contact

Email: shunyu.yao.physics@gmail.com

Linkedin: @shunyu yao