I'm an AI researcher working on personal AGI at OpenAI.
Previously, I was building AI for Optimus at Tesla. My contributions spanned the learning stack: data, deployment & evals, pre- & post-training, and sim ⇄ real.
My PhD at Stanford was advised by Chelsea Finn and Jiajun Wu. My thesis work focused on visual tokenization with an emphasis on disentanglement. Besides that, several of my projects tackled key ingredients of modern robot learning: effective data collection via wrist cameras and reliable sim evals.
I'm a bilingual Taiwanese-Canadian. In my free time, I like to ski/snowboard and play games, especially Soulslikes.