At Microsoft AI, I research infrastructure, data, and algorithms to improve RL numerics and training dynamics for reasoning (math and coding) and post-training, which has been the most enjoyable experience in my life.
Previously, I was a member of the Seed-Infra-Training framework team at ByteDance, where I built distributed training systems for multimodal and video generation models. I graduated from Shanghai Jiao Tong University in 2019 and
left the PhD program at UC Santa Barbara in 2022, a great place I will always cherish, where I met my wife.