Zangwei Zheng
郑奘巍, Ph.D.
AI Researcher & Full Stack Developer Tech Lead at HPC-AI Tech
Hi, I'm Zangwei👋, founder of Open-Sora and Video-Ocean. I'm working on AI-powered video generation and SaaS products that bridge research and real-world creativity.
I earned my Ph.D. from 
National University of Singapore (NUS), advised by Prof. Yang You. I had my research intern in 
Bytedance, mentored by Xuan Zhou and Youlong Cheng. I received my B.Sc. in Computer Science from 
Nanjing University (NJU). I worked remotely as a research intern at 
UC Berkeley for a year, mentored by Dr. Xiangyu Yue and Prof. Alberto L. Sangiovanni-Vincentelli. You can find more about my background in my CV.
🔬 Research Interests: Video Generation, Efficient ML and DL (Deep Learning) Optimization.
New homepage zangwei.dev launched! Welcome to visit!
New homepage zangwei.dev launched! Welcome to visit!
🎉 nextjs-portfolio-blog-research template for this site is also available on GitHub. On click and deploy your own homepage in minutes.
Video-Ocean Agent Online
Video-Ocean Agent Online
🌊 Video-Ocean Agent is now online! Generate videos up to minutes with a few clicks.
Ph.D. Graduated
Ph.D. Graduated
🎓 Graduated from HPC-AI Lab @ NUS. Thanks to my supervisor and friends' help! Congratulations to myself!
Open-Sora 2.0 
Released!
Open-Sora 2.0
🚀 Despite constrained computational resources, it achieves Sora-level metrics, fulfilling our goal of providing an open-source implementation of Sora.
Check out my latest work
Video-Ocean Video Agent
Video-Ocean Video Agent
Video-Ocean Video Agent generates videos up to minutes with a few clicks, including voice and face.
](https://github.com/hpcaitech/Open-Sora)](/_next/image?url=%2Fproj-open-sora.png&w=3840&q=75&dpl=dpl_HCUo1495g93x9UWRF8EtcovpMqzW)
Open-Sora
Open-Sora
The world's first open-source Sora-like video generation model — bringing efficient, high-quality video production to everyone.
](https://www.producthunt.com/products/video-ocean-v2-0)](/_next/image?url=%2Fproj-video-ocean.png&w=3840&q=75&dpl=dpl_HCUo1495g93x9UWRF8EtcovpMqzW)
Video-Ocean 
Video-Ocean
Video-Ocean is a video generation platform that allows you to generate videos, images and audios with state-of-the-art models.
](/_next/image?url=%2Fproj-cowclip.png&w=3840&q=75&dpl=dpl_HCUo1495g93x9UWRF8EtcovpMqzW)
CowClip Optimizer 
CowClip Optimizer
An optimizer that can train CTR prediction models with large batch (~128k)
Authors: Zangwei Zheng, Pengtai Xu, Xuan Zou, Da Tang, Zhen Li, Chenguang Xi, Peng Wu, Leqi Zou, Yijie Zhu, Ming Chen, Xiangzhuo Ding, Fuzhao Xue, Ziheng Qin, Youlong Cheng, Yang You
](/_next/image?url=%2Fproj-came.png&w=3840&q=75&dpl=dpl_HCUo1495g93x9UWRF8EtcovpMqzW)
CAME Optimizer 
CAME Optimizer
We introduce CAME, a confidence-guided optimizer that combines the fast convergence of adaptive methods with the low memory footprint of memory-efficient ones.
Authors: Yang Luo, Xiaozhe Ren, Zangwei Zheng, Zhuo Jiang, Xin Jiang, Yang You
MERIT Optimizer
MERIT Optimizer
MERIT constructs element-wise trust ratios to enable more robust update scaling by focusing on local weight structures.
Authors: Yang Luo, Zangwei Zheng, Ziheng Qin, Zirui Zhu, Yong Liu, Yang You
Dynamic Sequence Parallelism
Dynamic Sequence Parallelism
A novel, elegant and super efficient sequence parallelism for multi-dimensional transformer architecture.
Authors: Xuanlei Zhao, Shenggan Cheng, Chang Chen, Zangwei Zheng, Ziming Liu, Zheming Yang, Yang You
InfoBatch: Dataset Pruning on the Fly
InfoBatch: Dataset Pruning on the Fly
InfoBatch accelerates model training by pruning easy samples and reweighting losses, achieving up to 40% faster training without sacrificing accuracy across diverse tasks.
Authors: Ziheng Qin, Kai Wang, Zangwei Zheng, Jianyang Gu, Xiangyu Peng, Zhaopan Xu, Daquan Zhou, Lei Shang, Baigui Sun, Xuansong Xie, Yang You
](https://github.com/XueFuzhao/OpenMoE)](/_next/image?url=%2Fproj-openmoe.png&w=3840&q=75&dpl=dpl_HCUo1495g93x9UWRF8EtcovpMqzW)
HPC-AI Tech
Bytedance
UC Berkeley
Research Achievement Award of NUS
ACL 2023 Outstanding Paper Award
AAAI 2023 Distinguished Paper Award
Papers with Code Contributor Award
Outstanding Graduates of Nanjing University
CS5242: Neural Networks and Deep Learning
Algorithm Analysis & Design
National University of Singapore
National University of Singapore
From Video Generation to Commercial Agents
Tencent, invited by Kai Wang
Tencent, invited by Kai Wang
AI Video Gen Model From Scratch
Nvidia, invited by Yuyang Zhao
Nvidia, invited by Yuyang Zhao
Efficient Training with Open-Sora
National University of Singapore
National University of Singapore
Video Generation Model
Distributed AI (DAI) Conference, invited by Wenbin Li
Distributed AI (DAI) Conference, invited by Wenbin Li
Continual Learning on Pretrained Foundation Models
Get in Touch
Want to chat? Feel free to reach out
Explore collaboration opportunities
Want to chat? Feel free to reach out
- Ask questions
- Explore collaboration opportunities
- 15-minute coffee chat
Did you know?
A few interesting things about me
Academic Lineage
I am a 18th-generation academic descendant of Isaac Newton! From Newton to the present — a centuries-long mentorship chain spanning over 350 years.
Family Lineage
I am a 99th-generation descendant of You Ji (姬友), Duke Huan of Zheng (郑桓公) from ~800 BC! My family lineage spans over 2,800 years of Chinese history, from ancient Zhou dynasty to the present.