Bozheng Li 李泊铮
 |
Bozheng Li |
About me
I am currently a Research Scientist at ByteDance, working on VideoLLMs for Agentic Video Moderation. Previously, I received my master's degree from the Department of Computer Science at Brown University, fortunate to be advised by Prof. Chen Sun. Before that, I graduated from Zhejiang University with a B.Eng degree in Information Engineering, under the supervision of Prof. Yunlong Yu and worked closely with Dr. Mushui Liu.
I spent a wonderful summer at Harvard University as a research intern, privileged to work with Dr. Wanhua Li and Prof. Hanspeter Pfister. I also experienced the best journey at OpusClip working with Wenbo Zhu and Jiawang Cao, exploring video understanding in complex real-world tasks towards the agentic paradigm.
My research interests center on multimodal video understanding, from deploying models in product-level applications to enabling real-world spatial intelligence, with the long-term goal of building proactive video agents that can perceive, reason, and act across both the digital and physical worlds.
News
07/2026: Join ByteDance as a Research Scientist, working on VideoLLMs for Agentic Video Moderation.
09/2025: Agent Opus is released, best video editing agent ever.
06/2025: Start my internship at Harvard VCG Group.
06/2025: VEU-Bench got accepted by CVPR 2025 as highlight.
12/2024: TSAM got accepted by AAAI 2025.
10/2024: OAIR won the first place at the Hour-long VideoQA track of Second Percpetion Test Challenge @ ECCV2024
06/2024: MM-Screenplay won the first place at the Long-Term Video Question Answering track of LOVEU workshop @ CVPR2024
04/2024: Start my internship at OpusClip.
03/2024: Omni-CLIP is accepted by ECAI 2024.
Selected Projects
 |
PVD: Probing VideoLLMs for 3D Awareness |
 |
VEU-Bench: Towards comprehensive understanding of video editing |
 |
TSAM: A Temporal Sequence-Aware Model for Few-Shot Action Recognition |