GRAPHICS · MULTI-MODAL GENERATION · AGENT
I am a Researcher at TikTok, specializing in Multi-Modal Generation and Agent. Previously, I worked as a Senior Researcher at Tencent Youtu Lab and a Graphics Algorithm Engineer at Kwai. Prior to my full-time career, I completed research internships at Kwai (mentored by Haibin Huang and Chongyang Ma) and Tencent AI Lab. I received my B.S. and M.S. degrees from Huazhong University of Science and Technology (HUST) in 2018 and 2021, respectively, under the supervision of Prof. Nong Sang.
📢 I am looking for self-motivated interns to work on Multi-Modal Generation and related topics. If you are passionate about research and have strong coding skills, please drop me an email with your CV.
Research Journey
Stay Curious. Keep Exploring2020 → 2026- Hand3D Pose
- FaceSparse-view 3D; SDF
- ClothNeural Deformation
- HairReconstruction · Generation · Simulation
- 3D GenHair · Motion · Cloth
- VideoGenIdentity · Editing
- MM GenerationControllable Generation
Core Threads
3D Digital Humans · Controllable Generation · Multimodal Agents
News
- Our papers SteerVTE and Conformal Cache were accepted to NeurIPS 2026!
- Awarded 2nd Place ("XuanYuan" on Leaderboard) in ACM MM Identity-Preserving Video Generation (IPVG) Challenge.
Selected Publications
View All
Conformal Cache: Reliable Proxy-Discrepancy Caching for Fast Generative Inference
Professional Services
- Conference Reviewer: NeurIPS, ECCV, ICCV, AAAI, ACM MM, SIGGRAPH, Eurographics
- Journal Reviewer: TVC, TVCJ
Honors & Awards
- Outstanding Graduate, 2018 & 2021
- National Scholarship Top 2%, 2017
Education
- 2018.09 ~ 2021.07, HUST, AIA Dept, Master
- Supervisor: Prof. Nong Sang
- 2014.09 ~ 2018.07, HUST, OEI Dept and Qiming College, Bachelor