Jinwei Gu

Jinwei Gu is a computer vision researcher whose work runs from computational photography to world models for physical AI. He earned a PhD at Columbia University in 2010, taught at the Rochester Institute of Technology, and worked at Futurewei and at NVIDIA Research before leading R&D at SenseBrain from 2018 to 2023, where his team co-developed AI imaging sensors with Sony for flagship smartphones. A 2019 paper on neural RGB-D sensing, written with Chao Liu, Kihwan Kim, Srinivasa Narasimhan and Jan Kautz, was a CVPR best paper finalist. In March 2023 he joined the Chinese University of Hong Kong, and he coauthored NVIDIA's Cosmos world foundation model platform in 2025 and Cosmos 3 in 2026. He is a Principal Research Scientist and Senior Manager at NVIDIA's Cosmos Lab and an adjunct associate professor at CUHK.

Now at: NVIDIA and Chinese University of Hong Kong
Badges: Bridge

Timeline

2002–2010

Degrees from Tsinghua and Columbia

He earns a B.S. in 2002 and an M.S. in 2005 from Tsinghua University, then a PhD from Columbia University in 2010, awarded with distinction. In 2006, he coauthors "Time-varying Surface Appearance: Acquisition, Modeling and Rendering" for ACM Transactions on Graphics (SIGGRAPH) with Chien-I Tu, Ravi Ramamoorthi, Peter Belhumeur, Wojciech Matusik and Shree Nayar.
2010–2015

Assistant professor at RIT

From 2010 to 2013, he is an assistant professor in the Center for Imaging Science at Rochester Institute of Technology. From 2013 to 2015, he is a senior researcher in the America Media Lab at Futurewei Technologies.
2015–2018

Research scientist at NVIDIA

He is a Senior Research Scientist at NVIDIA Research from 2015 to 2018. He is a core contributor to the AI Co-Pilot SDK, which covers head pose, gaze estimation, drowsiness detection and gesture recognition and ships in the NVIDIA DRIVE IX SDK.
2018–2023

Executive Director at SenseBrain

From 2018 to 2023, he is R&D Executive Director at SenseBrain. He builds and leads a 30-person R&D team that co-develops AI imaging sensors with Sony for flagship smartphones.
2019

A CVPR best paper finalist

In 2019, he coauthors "Neural RGB->D Sensing: Depth and Uncertainty from a Video Camera" with Chao Liu, Kihwan Kim, Srinivasa Narasimhan and Jan Kautz. The CVPR 2019 oral paper uses deep learning to estimate per-pixel depth and its uncertainty from a monocular video stream, and it is a best paper finalist.
March 2023

Joining CUHK computer science

He joins the Department of Computer Science and Engineering at the Chinese University of Hong Kong in March 2023, with research on low-level computer vision, computational photography, computational imaging and appearance modeling. He is an associate editor of IEEE Transactions on Pattern Analysis and Machine Intelligence from 2022.
January 2025

The Cosmos platform paper

In January 2025, he is one of the authors of "Cosmos World Foundation Model Platform for Physical AI". The platform covers a video curation pipeline, pre-trained world foundation models, post-training examples and video tokenizers, released as open source with open-weight models.
As of October 2026

World models at NVIDIA Cosmos

He is a Principal Research Scientist and Senior Manager at NVIDIA's Cosmos Lab, where he co-leads research in world models for Physical AI, including Cosmos and Cosmos 3, and in computational imaging and camera systems. In June 2026, he is an author of Cosmos 3, a family of omnimodal world models that process and generate language, image, video, audio and action sequences. He is also an adjunct associate professor at CUHK.