Remi Cadene
Co-Founder And CEO at UMA
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolRemi Cadene is a founder and AI scientist who builds embodied intelligence—now leading UMA as Co-Founder and CEO after scaling real-world AI at Tesla Autopilot and launching foundational robot learning work like LeRobot at Hugging Face. With a PhD from Sorbonne and postdoctoral work at Brown, he blends rigorous research on neural architectures, explainability and human behavior modeling with production experience shipping neural nets for Tesla Optimus. He has 11+ years of hands-on ML engineering and MLOps, contributing widely to open-source projects such as pretrained-models.pytorch and the popular LeRobot effort that makes end-to-end robot learning more accessible. Remi’s background spans vision-and-language research at Facebook AI to core data and feature-engineering work in VQA systems, reflecting both theoretical depth and pragmatic system-building. Based in Paris, he draws inspiration from neuroscience and a commitment to open source to push robotics from lab demos toward everyday quality-of-life impact.
11 years of coding experience
6 years of employment as a software developer
Licence 3, Economics and Financial Management, Licence 3, Economics and Financial Management at Université Paris CitéLicence 1&2, Economics and Financial Management, Licence 1&2, Economics and Financial Management at Paris-Sorbonne University Abu DhabiDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Sorbonne UniversityMaster's degree, Computer Science, Master's degree, Computer Science at Pierre and Marie Curie University
French, English, German
Stackoverflow
57reputation
5kreached
4answers
4questions
Github Skills (28)
pytorch10
convolutional-neural-networks10
robotics10
py10
python10
pre-trained-model10
imagenet10
setuptools10
machine-learning10
io10
reinforcement-learning10
deep-learning10
resnet10
data-processing10
neural-network10
Programming languages (10)
Github contributions (5)
Cadene/pretrained-models.pytorch
Apr 2017 - Apr 2019
Pretrained ConvNets for pytorch: NASNet, ResNeXt, ResNet, InceptionV4, InceptionResnetV2, Xception, DPN, etc.
Role in this project:
ML Engineer
ML EngineerContributions:117 commits, 26 PRs, 90 pushes in 2 years
Contributions summary:Remi primarily contributed to the development and implementation of pretrained models within the PyTorch framework. They added support for various ResNeXt architectures, Inception-based models (including InceptionResNetV2 and InceptionV4), and DualPathNetworks. They also improved the API for several models and introduced a toy-example demonstrating the usage of these pretrained models, indicating a focus on model integration and usability. Moreover, the user added support for pre-trained weights of NASNet-A-Large, further demonstrating proficiency in integrating and utilizing advanced convolutional neural network architectures.
pytorchresnetxceptionimagenetresnext
Cadene/vqa.pytorch
Apr 2017 - Dec 2019
Visual Question Answering in Pytorch
Role in this project:Back-end & MLOps Engineer
Back-end & MLOps EngineerContributions:1 release, 31 commits, 8 PRs in 2 years 8 months
Contributions summary:Remi primarily contributed to the project by modifying core functionalities of the Visual Question Answering (VQA) system built in PyTorch. Their changes involved fixing data loading and preprocessing, adding logging and configuration management features, and integrating VQA2 and Visual Genome datasets. They also adapted the extraction script to extract features, showcasing a focus on model training and data handling, with elements of MLOps practices in the logging and configuration setup.
pytorchvisual-question-answeringvqadeep-learningresnet
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.