Robin Strudel
Research Scientist at Inria
Paris, Ile-de-France
Summary
👤
Senior🎓
Top SchoolRobin Strudel is a Research Scientist at DeepMind Paris with eight years of experience building and evaluating generative models and vision systems rooted in strong mathematical foundations. He holds a PhD from INRIA and École Normale Supérieure, where his Willow-team work bridged learning methods and visually guided robotic tasks under advisors Ivan Laptev and Cordelia Schmid. His background spans probability, PDEs, and stochastic modeling from stints at Oxford and UC Berkeley, giving him a rare ability to connect theoretical limits to practical ML systems. Robin actively contributes to open-source projects—improving semantic segmentation tooling (Segmenter) and validating robotics dynamics code (Pinocchio)—demonstrating attention to both research quality and engineering robustness. Based in Paris, he blends rigorous academic training with hands-on engineering at scale, and maintains a public portfolio at rstrudel.github.io.
8 years of coding experience
1 year of employment as a software developer
Master's degree, Probability and PDE, Master's degree, Probability and PDE at Ecole normale supérieure de LyonLycée Louis Le GrandMaster's degree, MVA, Master's degree, MVA at École Normale Supérieure Paris-Saclay
French, English
Github Skills (21)
pytorch10
python10
testing10
machine-learning10
inference10
image-segmentation10
segmentation10
trainings10
computer-vision10
transformer10
modeling10
c-language9
deep-learning9
cprogramming-language9
automatic-differentiation3
Programming languages (4)
Github contributions (5)
rstrudel/segmenter
May 2021 - Mar 2022
[ICCV2021] Official PyTorch implementation of Segmenter: Transformer for Semantic Segmentation
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 54 commits, 4 PRs in 9 months
Contributions summary:Robin's primary contribution involves modifying and adding to the core training and inference scripts of the Segmenter model. They updated the inference script to incorporate a progress bar and set the model to run on the GPU. Additional changes include setting up checkpoint loading for evaluation and integrating the use of checkpoint resolution. The user implemented modifications to the metrics computation and simplified the patch creation process.
pytorchsemantic-segmentationsegmentertransformericcv2021
stack-of-tasks/pinocchio
Jan 2020 - Jan 2020
A fast and flexible implementation of Rigid Body Dynamics algorithms and their analytical derivatives
Role in this project:QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:8 commits, 3 PRs, 9 comments in 21 days
Contributions summary:Robin's contributions primarily involve adding and modifying example scripts within the repository. Their work focuses on verifying the functionality of the `pinocchio` library, specifically related to collision detection, mesh loading, and capsule approximation. The user implemented tests to display shapes, incorporate mesh scaling, and ensure the constraints used in capsule approximation are correctly enforced, including adding assertion checks. This involved making changes to existing examples and modifying existing code to perform additional testing steps.
code-generationpythonautomatic-differentiationc-plus-pluspinocchio
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.