Kevin (Qi) Zhao

About Me

Hi there, I am Kevin. Currently, I am an AI researcher at ByteDance Seed focusing on video generation and world model. I worked with Dr. Lu Jiang in video generation projects such as Seaweed, SeeDance, and Bagel2. I currently work with Dr. Yi Jiang on SeeDance2.0 Pro, focusing on long video generation and Haoqi Fan on SeeWorld, interactive world model.

Previously, I obtained my Master of Science degree at Brown University, where I was fortunate to collaborate closely with professor Chen Sun and his group on video understanding. I have also worked with professor George Konidaris and his PhD student Haotian Fu on embodied agents. I obtained my Bachelor of Science degree at NYU.

I grew up in Shanghai until I went to high school in Scottsdale, AZ.

Selected Publications

Context Unrolling in Omni Models

Team Omni Model, Qi Zhao, et al.

Tech Report

arXiv Website

Synthetic Video Enhances Physical Fidelity in Video Synthesis

Qi Zhao, Xingyu Ni, Ziyu Wang, Feng Cheng, Ziyan Yang, Lu Jiang*, Bohan Wang*

ICCV 2025

arXiv Website HuggingFace

SeeDance: Conditional Human Image Animation with Discrete Motion Representations

Team SeeDance, Qi Zhao, et al.

Technical Report

Website

Seaweed-7B: Cost-Effective Training of Video Generation Foundation Model

Team Seaweed, Qi Zhao, et al.

Technical Report

Website Paper

EPO: Hierarchical LLM Agents with Environment Preference Optimization

Qi Zhao*, Haotian Fu*, Chen Sun, George Konidaris

EMNLP 2024 Main

arXiv Code

Vamos: Versatile Action Models for Video Understanding

Shijie Wang, Qi Zhao, Minh Do, Nakul Agarwal, Kwonjoon Lee, Chen Sun

ECCV 2024

Website arXiv Code

AntGPT: Can Large Language Model Help Long-term Action Anticipation from Videos?

Qi Zhao*, Shijie Wang*, Ce Zhang, Changcheng Fu, Minh Do, Nakul Agarwal, Kwonjoon Lee, Chen Sun

ICLR 2024

Website arXiv Video Code