YaGuang Li (李亚光)

Principal Research Scientist, Google DeepMind

@Email Scholar LinkedIn
Large Language Models Post-training and RL Scaling Deep Learning on Graphs Spatiotemporal Forecasting
YaGuang Li

Biography

YaGuang Li is a Principal Research Scientist at Google DeepMind, working on Gemini DeepThink and the Gemini reasoning stack, with a focus on post-training and efficient reinforcement learning (RL) scaling. He co-led the finetuning and post-training efforts of Gemini 1.5 and Gemini 1.0 for Gemini Advanced. He also contributed to LaMDA and PaLM-2 across pretraining, instruction tuning, and efficient serving. He earned his Ph.D. in Computer Science from the University of Southern California, advised by Prof. Cyrus Shahabi and Prof. Yan Liu. His research spans large language models and deep learning on graphs, with applications in spatiotemporal forecasting and relational inference.

Experience

Selected Publications

Technical Reports and Preprints

Refereed Papers

Patents and Applications