My research focuses on robot video generation and world models, multimodal large language models, and agent systems. I am interested in building intelligent systems that can understand, predict, and reason about the physical world across modalities.
News
Experience
Research Intern, Tencent IEG
May 2026 - Late Sep. 2026
Research Intern, Tencent CSIG
Aug. 2025 - May. 2026
Research Intern, Tencent TEG
Aug. 2023 - Jul. 2025Research Collaboration
Weekly Meetings
Research Collaborator — Video Generation
Jun. 2026 - PresentRecent collaboration: ReCAP, efficient context scheduling for long-horizon video world models.
Research Collaborator — Spatiotemporal Predictive Learning
Dec. 2023 - Jun. 2025Close collaboration on physical spatiotemporal forecasting, including BeamVQ, a metric-guided latent-search framework.
Selected Publications
Google Scholar
RoboAlign-R1: Distilled Multimodal Reward Alignment for Robot Video World Models
Metric-Guided Latent Search for Physical Spatiotemporal Forecasting
Service
- Reviewer: ICLR, KDD, NeurIPS, ICCV, AAAI, TKDE, ICML, and ACM MM
- Research Areas: Multimodal large models, video world models, and agent systems
Selected Awards & Honors
Interests
Basketball
I make time for one to two hours of basketball most days. If you’re free, you’re always welcome to join me for a game!
My favorite moves are fadeaway jumpers and three-pointers. I’ve spent many happy hours playing at USTC’s Gaoxin Campus gym, the basketball gym at Tencent’s Beijing headquarters, and Tsinghua University’s Zijing courts.