I grew up in Hong Kong. Currently in Palo Alto, on leave from M&T at UPenn.
I work on Grok Model Capabilities at xAI.
I've spent time dabbling across corners of ML. Lately on scaling RL for code generation through synthetic perturbations, and turning LoRA adapters into dynamic mixtures of experts in vision models.
Earlier projects explored interpretability of truth in language models, biologically-inspired memory for agents, learned downsampling for 3D medical imaging, and graph networks for probabilistic reasoning.
I am also fascinated by venture investing. Trying to understand the technologies shaping the frontier and what it takes to push them forward. Interests range across AI/ML, fintech, crypto, robotics, biotech, and prediction markets. When I'm not backing friends building what I wish existed, find me at Dorm Room Fund.
Previously at Ramp, Nous Research, and KKR.
LinkedIn | GitHub | X/Twitter