Ming-Yu Liu
Vice President of Cosmos Lab at NVIDIA; supervised the Cosmos3 world model family
Ming-Yu Liu is Vice President of Cosmos Lab at NVIDIA and an IEEE Fellow, leading NVIDIA Cosmos, which builds world foundation models for physical AI. Liu is the sole person credited under "Supervision" in the Cosmos3 technical report, which introduced the Cosmos3-Nano and Cosmos3-Super omnimodal world models. Earlier generative work from Liu's group includes GauGAN, the research behind NVIDIA Canvas.124
Career
Liu is Vice President of Cosmos Lab at NVIDIA. NVIDIA Research's profile describes Liu as a Vice President of Research who leads a research group on deep generative models for content creation, and credits NVIDIA Cosmos, NVIDIA Canvas (GauGAN) and NVIDIA Maxine (LivePortrait) as products enabled by the group's research.12
Liu's site lists NVIDIA Pioneer Research Awards in 2017 and 2018, an R&D 100 Award in 2014, and a 2016 award from Mitsubishi Electric Research Labs (MERL), alongside SIGGRAPH Real-Time Live awards in 2019 and 2021.1
Research
Liu's earlier work on image synthesis includes "Unsupervised Image-to-Image Translation Networks" (2017) with Thomas Breuel and Jan Kautz, high-resolution image synthesis with conditional GANs (2017), and spatially-adaptive normalization for semantic image synthesis (2019), the GauGAN paper.678
Liu's group now builds world foundation models for physical AI. NVIDIA published the Cosmos World Foundation Model Platform paper in January 2025, and the June 2026 Cosmos3 report describes a family of omnimodal world models that process and generate language, image, video, audio and action, released with code and checkpoints under the OpenMDW-1.1 license.543
Work
NVIDIA Cosmos3-NanoOpen weights
Credited under Supervision in the June 2026 Cosmos3 technical report, which releases Cosmos3-Nano as an open-weight checkpoint.4
NVIDIA Cosmos3-SuperOpen weights
Credited under Supervision, and among the model architecture contributors, in the June 2026 Cosmos3 technical report, which releases Cosmos3-Super as an open-weight checkpoint.4
Semantic Image Synthesis with Spatially-Adaptive Normalization (GauGAN) (external site: arxiv.org)Published research
Co-author of the 2019 paper whose research NVIDIA credits as the basis of NVIDIA Canvas (GauGAN).82
Career timeline
2014
Received an R&D 100 Award from R&D Magazine.1
Mar 2017
Posted "Unsupervised Image-to-Image Translation Networks" as first author.6
Nov 2017
Co-authored "High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs".7
Mar 2019
Co-authored the spatially-adaptive normalization (GauGAN) paper on semantic image synthesis.8
Jan 2025
Co-authored NVIDIA's Cosmos World Foundation Model Platform for Physical AI paper.52
Jun 22, 2026
Supervised the Cosmos3 omnimodal world models described in NVIDIA's technical report.4
Honors and recognition
Related people
More sourced profiles at the same organization, in the same field, or from the same school.
- Aaron LefohnAlso at NVIDIA
- Aditya RameshImage, video, and 3D generation
- Azita MartinAlso at NVIDIA
- Alex NicholImage, video, and 3D generation
- Bill DallyAlso at NVIDIA
- Alexei A. EfrosComputer vision
- Boris GinsburgAlso at NVIDIA
- Amir HertzImage, video, and 3D generation
- Bryan CatanzaroAlso at NVIDIA
- Amnon ShashuaComputer vision
- Dan SuAlso at NVIDIA
- Angjoo KanazawaComputer vision
Is this you?
Profiles cover public professional work only, and every line cites a source. If something is out of date, missing or wrong, send a correction with a link to an official page and we will update it. Suggest a correction
Add this badge to your website or GitHub profile README. It links back to this page. It is not an endorsement or a rating.
Markdown (for a README)
[](https://unitedstatesofamericasuperintelligence.com/people/ming-yu-liu/)
HTML
<a href="https://unitedstatesofamericasuperintelligence.com/people/ming-yu-liu/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/profile.svg" alt="Ming-Yu Liu — Profile on USASI" height="20"></a>
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.