- VP of Research, Google DeepMind21
Clément Farabet is VP of Research at Google DeepMind, the title on Farabet's bylines for Google's March 2025 Gemma 3 and April 2026 Gemma 4 launch posts; the Gemma 4 post says those models are released under a commercially permissive Apache 2.0 license. Farabet is listed among the core contributors of the Gemma 4 Technical Report (2026) and among the sponsors of the earlier Gemma, Gemma 2 and Gemma 3 reports. A 2022 NVIDIA blog post describes Farabet as NVIDIA's vice president of AI infrastructure, designing a data factory for autonomous-vehicle development on the SaturnV supercomputer. Farabet is also named, with NYU, among the copyright holders of the Torch7 machine learning library.12345678
At a glance
- Current role
- VP of Research
- Company
- Google DeepMind
- Sector
- Company
- Work listed
- 2
- Sources cited
- 12
- Profile reviewed
Research and open-source work
Farabet's early research, with Camille Couprie, Laurent Najman and Yann LeCun, applied multiscale convolutional networks to scene parsing, labeling every pixel of an image with the category of the object it belongs to. A 2013 follow-up extended the approach to indoor scenes using depth information.910
Farabet is named with NYU among the copyright holders of Torch7, a machine learning library whose C backend libraries were later extended and rewritten in the ATen C++11 library. Farabet's GitHub profile also hosts companion projects such as torch-ios, a port of Torch7 to iOS, and gfx.js, a browser-based graphics backend with a Torch7 client.81112
Career
In a 2022 NVIDIA blog post, Farabet was introduced as NVIDIA's vice president of AI infrastructure, with experience that included building Twitter's AI machine. At NVIDIA, Farabet designed a scalable data factory on the SaturnV supercomputer for training neural networks used in autonomous-vehicle development.7
At Google DeepMind, Farabet holds the title of VP of Research and has co-signed the launch posts for Gemma 3 and Gemma 4. Farabet appears among the sponsors of the Gemma, Gemma 2 and Gemma 3 technical reports and among the core contributors of the Gemma 4 report.124563
Work and contributions
Gemma Open weights
Co-author of Google's Gemma 3 (2025) and Gemma 4 (2026) launch posts, listed among the core contributors of the Gemma 4 Technical Report, and among the sponsors of the Gemma, Gemma 2 and Gemma 3 technical reports; the Gemma 4 post says the models are released under Apache 2.0.123456
Torch7 Open-source code
Named as a copyright holder (NYU, 2011–2013) in Torch7's BSD-style COPYRIGHT file; Farabet's GitHub profile hosts related packages such as torch-ios (Torch7 for iOS) and gfx.js, a graphics backend with a Torch7 client.812
Timeline
- Posted "Scene Parsing with Multiscale Feature Learning, Purity Trees, and Optimal Covers" as first author.9
- Co-authored "Indoor Semantic Segmentation using depth information" with Camille Couprie, Laurent Najman and Yann LeCun.10
- Was introduced by NVIDIA as its vice president of AI infrastructure.7
- Listed among the sponsors of the first Gemma technical report.4
- Co-authored Google's launch post for Gemma 3 as VP of Research at Google DeepMind.1
- Co-authored Google's launch post for Gemma 4, released under the Apache 2.0 license.2
- Listed among the core contributors of the Gemma 4 Technical Report.3
Other documented roles
Roles the sources document that are not confirmed as current.
Where they work
- Google DeepMind 64 people in the wiki
Related people
- Laurent SartranAlso at Google DeepMind
- Danqi ChenAlso at Google DeepMind
- Stephen RollerAlso at Google DeepMind
- Arthur MenschAlso at Google DeepMind
- Noam ShazeerAlso at Google DeepMind
- Jeff DeanTraining systems · ML frameworks and libraries
- Quoc LeLanguage models · Training systems
- Sara HookerLanguage models · Training systems
- Nils ReimersLanguage models · ML frameworks and libraries
- Edward BeechingLanguage models · Training systems
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
12 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026
- Introducing Gemma 3 (Google blog launch post, byline VP of Research, Google DeepMind)
- Gemma 4 (Google blog launch post, byline VP of Research, Google DeepMind; Apache 2.0 license)
- Gemma 4 Technical Report (arXiv 2607.02770), core contributor list
- Gemma: Open Models Based on Gemini Research and Technology (arXiv 2403.08295), contributor list
- Gemma 2: Improving Open Language Models at a Practical Size (arXiv 2408.00118), contributor list
- Gemma 3 Technical Report (arXiv 2503.19786), contributor list
- Music to the Gears: NVIDIA's Clément Farabet on Orchestrating AI Training for Autonomous Vehicles (NVIDIA blog)
- torch/torch7 COPYRIGHT.txt (BSD-style license and copyright holders)
- Scene Parsing with Multiscale Feature Learning, Purity Trees, and Optimal Covers (arXiv 1202.2160)
- Indoor Semantic Segmentation using depth information (arXiv 1301.3572)
- torch/torch7 README (development status)
- clementfarabet (Clement Farabet) on GitHub