I am currently a Research Scientist at Intelligent Creation Lab, ByteDance.

Before that, I worked at Alibaba DAMO/Tongyi Lab, and graduated from Tsinghua University in 2020. My research interests focus on computer vision and multimodal systems, specifically on human-related perception and generation. I have served as a conference reviewer for NeurIPS, ICLR, AAAI, ICML, CVPR, ICCV, ECCV, etc.

My research on human-related perception has received 7 winner awards in CVPR challenges and was included in the Stanford 2022 AI Index Report, and was also applied in the 2022 Beijing Winter Olympics. I also had research interests in training with noisy data: NGC, RLIP, RLIPv2, LOOK, ViM, GMNet and etc.

Recently, I have been exploring research on highly realistic avatar technology and its applications in film production. I am leading a unified human video generation project, including Loopy, CyberHost and OmniHuman-1, which has been covered by media such as Forbes, ABC, Wired, Business Insider, VentureBeat, CNR News, 机器之心 and others.