Ming-Yu Liu (NVIDIA)

Liu received his bachelor’s degree from National Chiao Tung University in 2003.⁠[1]

At the Department of Electrical and Computer Engineering of the University of Maryland, College Park, he received a master’s degree in 2011 and a PhD in 2012; his doctoral adviser was Rama Chellappa.⁠[1][1][2]

Before joining NVIDIA, he was a principal research scientist at Mitsubishi Electric Research Laboratories (MERL) in Cambridge, Massachusetts, and had also worked in software engineering at Intel. While at MERL, he received a 2014 R&D 100 Award for his contribution to a commercial robotic bin-picking system, his street-view semantic labeling paper was a best paper finalist at the 2015 robotics conference RSS, and he received MERL’s CR&D Award in 2016. In June 2016, he and Oncel Tuzel published the paper on coupled generative adversarial networks (CoGAN).⁠[1][2][2][3][2][4][5][5]

In 2016, Liu joined NVIDIA Research, where his research focus was generative image modeling, with the goal of giving machines a human-like capacity for imagination. He received NVIDIA’s internal Pioneer Research Award twice, in 2017 and 2018.⁠[3][3][4]

With five co-authors, he published the paper “High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs,” known as pix2pixHD, which uses conditional generative adversarial networks to synthesize high-resolution images from semantic label maps.⁠[6][6][3]

In 2018, he won first place in both the Domain Adaptation for Semantic Segmentation Competition and the Robust Optical Flow Estimation Challenge held at CVPR, and was named an outstanding reviewer for CVPR 2018.⁠[2][4]

In March 2019, he and three co-authors led by first author Taesung Park published the paper on semantic image synthesis with spatially-adaptive normalization (SPADE), which was a best paper finalist at CVPR 2019. The GauGAN app built on this research lets users paint shapes and has a neural network fill in photorealistic details; it won both the Best in Show Award and the Audience Choice Award at SIGGRAPH 2019 Real-Time Live!, and also received Popular Science’s 2019 “Best of What’s New” Award. He explained that the project’s starting point was simply to improve their previous-generation image synthesis algorithm, pix2pixHD, and that they had not expected many people to find the preliminary results already useful.⁠[7][7][2][1][2][3][3]

His work on generative adversarial networks for video compression helped NVIDIA win the “Most Disruptive Inventor” award in ForbesForbesForbes is an American business media brand that began with Forbes magazine, founded by B.C. Forbes in 1917. It is known for rankings such as the Forbes 400, the World’s Billionaires list and 30 Under 30, and operates Forbes.com, ForbesLive events and licensed editions in 81 countries. Since 2014 a majority stake has been held by the Hong Kong investment group Integrated Whale Media; a 2021 plan to list on the New York Stock Exchange through a special purpose acquisition company was terminated in 2022. It is headquartered in Jersey City, New Jersey.Open the full entry’ 2020 AI awards; in 2021, he again won the Best in Show Award at SIGGRAPH Real-Time Live!.⁠[2][4][4]

In 2023, the University of Maryland’s Department of Electrical and Computer Engineering gave him its Distinguished Alumni Award; at the time he was a senior director of research at NVIDIA. He contributed to NVIDIA Picasso, NVIDIA Maxine, and NVIDIA Canvas: for Picasso he built a generative AI foundry that helps customers create image, video, and 3D generators trained with responsible data sources, and his team developed a live portrait system for Maxine to enhance video communication. The same year, Neuralangelo, a 3D surface reconstruction project he worked on, was named one of TIMETIMETIME is an American newsmagazine and media brand founded by Henry Luce and Briton Hadden, which published its first issue on March 3, 1923; it is headquartered in New York and known for its Person of the Year selection and the TIME100 list. Since 2018 it has been personally owned by Marc and Lynne Benioff; its current Chief Executive Officer is Jessica Sibley, and its Editor-in-Chief is Sam Jacobs.Open the full entry’s Best Inventions of 2023.⁠[1][1][1][1][8][8][4]

NVIDIA launched Cosmos, a platform for physical AI development, at CES, and Liu wrote about it: Cosmos is a family of world foundation models that can predict and generate physics-aware videos of the future state of a virtual environment, helping developers build next-generation robots and autonomous vehicles; the first models were made available under an open model license that allows commercial use. Cosmos won Best AI and Best Overall in the Best of CES Awards presented by the CNET Group. The next day, the NVIDIA team submitted the technical report “Cosmos World Foundation Model Platform for Physical AI” to arXiv.⁠[9][9][9][9][9][10][10]

In 2026, he received a Distinguished Alumni Award from the College of Electrical and Computer Engineering (ECE College) of Chiao Tung University.⁠[4][4]

NVIDIA announced the world foundation model Cosmos 3 at NVIDIA GTC Taipei during COMPUTEX, and Liu wrote about it: the model combines vision reasoning with multimodal generation across text, video, images, ambient sound, and action in a single model; it uses a mixture-of-transformers architecture in which a reasoning block first interprets a scene and a generation block then produces synthetic video or robot-task data, and it can natively generate numerical action data such as joint angles and gripper positions.⁠[11][11][11][11][11]

At the Runway AIRunwayRunway (Runway AI, Inc.) is a New York–based artificial intelligence company that grew out of the ITP program at New York University’s Tisch School of the Arts in 2018. It develops video generation models and tools for the film, advertising and creative industries, has released the Gen-1 through Gen-4.5 series of models, and has extended its research to “general world models” that simulate the world in real time.Open the full entry Summit in San Francisco, he joined Google DeepMindGoogle DeepMindGoogle DeepMind is Google’s artificial intelligence research organization, formed in April 2023 by merging DeepMind, which Google acquired in 2014, with the Brain team from Google Research, and led by CEO Demis Hassabis. It developed AlphaGo, AlphaFold, and the Gemini family of models, and Hassabis and John Jumper received the 2024 Nobel Prize in Chemistry for AlphaFold’s protein structure prediction work.Open the full entry principal research scientist Jon BarronJon Barron (Google DeepMind)Jon Barron is an American computer vision researcher and a principal research scientist at Google DeepMind, where he leads a small team in San Francisco that mostly works on neural radiance fields (NeRF). He is one of the authors of the 2020 NeRF paper and, as first author, published follow-up work including mip-NeRF, mip-NeRF 360, and Zip-NeRF.Open the full entry and Wayve research director Alexander Toshev, as vice president of NVIDIA’s Cosmos Lab, on the panel “Grounding Intelligence in the Physical World,” discussing how to close the gap between simulation and the real world and where world models go next.⁠[12][12][12]

Liu has said that the world foundation models developed by Cosmos are meant to let machines not just perceive the world but also simulate, reason about, and interact with it. He describes his vision as building the “MatrixMatrix PartnersMatrix Partners, now branded Matrix, is a US early-stage venture capital firm with roots in Hellman, Ferri Investment Associates, established in 1977. Matrix was founded in 1982 and invests in technology companies from the idea stage through Series A.Open the full entry” for robots: a physics-grounded simulated universe where machines can “dream,” rehearse, and accumulate experience safely at scale before ever acting in the real world. Three products, Cosmos, NVIDIA Canvas (GauGAN), and NVIDIA Maxine (LivePortrait), stem from his research group’s work.⁠[4][4][4][8]

Personal website: https://mingyuliu.net/ (opens in a new window)⁠[4]

X: Public account: https://x.com/liu_mingyu (opens in a new window)⁠[4]

LinkedInLinkedInLinkedIn is a social networking platform for professionals. It was founded in 2002 by Reid Hoffman and others, launched in 2003, and listed on the New York Stock Exchange in 2011; after Microsoft acquired it for approximately $26.2 billion in 2016, it became a wholly owned Microsoft subsidiary. Its revenue comes from membership subscriptions, advertising, and recruitment solutions.Open the full entry: Public profile: https://www.linkedin.com/in/mingyuliu/ (opens in a new window)⁠[4]

NVIDIA: Official website: https://www.nvidia.com/ (opens in a new window).⁠[4]