Yue Cao
Yue Cao is a computer-vision researcher whose work spans Swin Transformer and video-generation models. He earned bachelor's and doctoral degrees at Tsinghua and completed his PhD in 2019. He worked as a senior researcher at Microsoft Research Asia and coauthored Swin Transformer, which won ICCV's 2021 Marr Prize for Best Paper. He joined BAAI and co-founded Lightyears Away before he founded Sand.ai in 2024; in 2025, he coauthored the MAGI-1 world model. As of October 2026, he is CEO of Sand.ai.
Timeline
-
2014–2019
A PhD at Tsinghua
He earns bachelor's and doctoral degrees at Tsinghua and completes his PhD in 2019 under Jianmin Wang's co-supervision.
-
2019–2022
Research at Microsoft Asia
He works as a senior researcher at Microsoft Research Asia from 2019 to 2022.
-
2021
The Marr Prize
He coauthors Swin Transformer, a hierarchical vision Transformer with shifted windows. The paper receives ICCV's 2021 Marr Prize for Best Paper.
-
2022–2023
At BAAI and Lightyears Away
He joins BAAI in 2022 and becomes a co-founder of Lightyears Away in 2023.
-
January 2024
Founding Sand.ai
He founds Sand.ai in January 2024 to develop video-generation models.
-
2025
The MAGI-1 model
In 2025, he coauthors MAGI-1, a world model that generates videos by predicting fixed-length chunks autoregressively.
-
April 2025
Releasing MAGI-1 weights
Sand.ai releases the MAGI-1 model weights and inference code in April 2025.
-
As of October 2026
CEO of Sand.ai
As of October 2026, he is CEO of Sand.ai.
Career
- Since 2024Sand.aiFounder, CEO
- From 2023Lightyears AwayCo-founder
- From 2022Beijing Academy of Artificial Intelligence
- 2021–2023University of British Columbia
- 2021–2023University of British Columbia, Okanagan Campus
- 2020–2022Microsoft
- 2019–2022Microsoft ResearchSenior Researcher
Education
- 2019Tsinghua UniversityBachelor's degree
Most-cited works
- Swin Transformer: Hierarchical Vision Transformer using Shifted WindowsWith Han Hu, Stephen Lin and Yixuan Wei
- Swin Transformer V2: Scaling Up Capacity and ResolutionWith Furu Wei, Han Hu, Yixuan Wei and Zhenda Xie
- GCNet: Non-Local Networks Meet Squeeze-Excitation Networks and BeyondWith Han Hu and Stephen Lin
- Video Swin TransformerWith Han Hu, Stephen Lin and Yixuan Wei
- SimMIM: a Simple Framework for Masked Image ModelingWith Han Hu and Zhenda Xie
Sources and credits
Photo: We have no published photo of Yue Cao yet. Send us a free-licensed photo.
- Mingsheng Long - Tsinghua University
- ICPR22-SSL
- ICCV 2021 PAPER AWARDS | ICCV 2021
- Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows
- MaHui Entrepreneur | Sand.ai's Yue Cao: A Decade from Researcher to Entrepreneur Traces the Evolution of Large Models
- Exclusive: Sand.ai's Cao Yue Secures Over USD 100M in Financing - Why Video Is the Most Critical Path to the World Model
- MAGI-1: Autoregressive Video Generation at Scale
- sand-ai/MAGI-1 · Hugging Face
- Magi: AI Video Generator & AI Video Extender By Sand.ai


