Xuefeng Hu is a Member of Technical Staff at AMI Labs, where he conducts world-model research. His broader interests span computer vision, multimodal learning, representation learning and source-free domain adaptation, with an emphasis on building systems that can understand, generate and reason across visual information.

Before joining AMI Labs, Hu spent more than two years at ByteDance, progressing to Staff Research Scientist and leading video foundation-model pretraining. He contributed to unified models for multimodal understanding and generation, developed reinforcement-learning infrastructure for agentic multimodal systems, and worked on large-scale training pipelines encompassing pretraining, supervised fine-tuning and preference alignment. His earlier research experience included internships at Meta, Facebook and Amazon Lab126.

Hu earned a PhD in computer science from the University of Southern California under Professor Ram Nevatia, following a master’s degree from Columbia University and bachelor’s degrees from the University of Michigan and Shanghai Jiao Tong University. His research on refining vision-language models through source-free domain adaptation was named a WACV 2024 Best Paper Finalist. Outside the laboratory, he founded LA Stories, a nonprofit forum promoting interdisciplinary discussion and cultural life within Los Angeles’s Mandarin-speaking community.