Clément Farabet
VP, Research at Google DeepMind
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolClément Farabet is a veteran AI research and engineering leader with 16 years building foundational ML software and teams across DeepMind, NVIDIA, Twitter, and Google DeepMind where he now serves as VP of Research. He combines deep academic roots (PhD in Computer Science) with practical systems expertise, having co-created and contributed heavily to Torch7 and its CUDA backends—work that helped shape modern GPU-accelerated deep learning tooling. As an entrepreneur and co-founder of MadBits and Twitter Cortex, he has a proven track record of turning research into product-scale ML systems for computer vision and recommendation. At NVIDIA he led large-scale AI and AV infrastructure projects like MagLev and RAPIDS, bridging deep learning with classical data science. Known for hands-on low-level contributions (autograd, cutorch, nn modules) and large program leadership, he uniquely spans algorithmic innovation to production GPU systems. Based in Palo Alto, he continues to hire and mentor multidisciplinary teams focused on next‑generation AI software paradigms.
16 years of coding experience
10 years of employment as a software developer
Doctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at Université Paris-EstM.Eng, Electrical Engineering, M.Eng, Electrical Engineering at Institut National des Sciences Appliquées de Lyon
Github Skills (36)
thrust10
pytorch10
convolutional-neural-networks10
operation10
back-end-development10
image-processing10
package-development10
testing10
plugin-development10
tensorrt10
gpu-programming10
machine-learning10
c1110
autoencoder10
c1710
Programming languages (8)
Github contributions (5)
torch/demos
Jul 2011 - Nov 2013
Demos and tutorials around Torch7.
Role in this project:
ML Engineer
ML EngineerContributions:167 commits in 2 years 5 months
Contributions summary:Clément's contributions primarily focused on the development and enhancement of machine-learning related projects within the repository. They implemented various autoencoder models, including linear, PSD (predictive sparse decomposition), and convolutional variants. The user also integrated and modified the loss functions used, switching from MSE to ClassNLLCriterion. Finally, the user incorporated a new dataset for house numbers.
torch/tutorials
Jul 2012 - Feb 2014
A series of machine learning tutorials for Torch7
Role in this project:ML Engineer
ML EngineerContributions:48 commits in 1 year 7 months
Contributions summary:Clément primarily contributed to machine learning tutorials within the Torch7 framework, focusing on CUDA integration, dataset loading, and model adjustments. Their work involved optimizing code for GPU utilization, implementing dataset loading for CIFAR-10, SVHN, and MNIST, and adding support for MSE loss and tanh activations, along with the k-means algorithm. The commits demonstrate a strong understanding of machine learning concepts and their practical implementation within the Torch7 ecosystem.
deep-learningmachine-learning-tutorialsmachine-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.