Matteo Hessel
Senior Staff Research Engineering Manager at Google DeepMind
Milan, Lombardy, Italy
Summary
🤩
Rockstar🎓
Top SchoolMatteo Hessel is a Senior Staff Research Engineering Manager at Google DeepMind in Milan, leading engineering and research work on Gemini with a focus on agents and reasoning. He combines 11 years of deep learning and reinforcement learning experience with a prolific research record—30+ papers across top venues and 15,000+ citations—and multiple patents. Matteo is the lead developer behind widely used open-source ML libraries (Optax, Rlax, TRFL, Chex) and has contributed optimizers and numerical-stability improvements to prominent projects like JAX and TensorFlow Probability. He blends hands-on ML engineering (implementing novel optimizers, gradient-noise techniques and activation functions) with people leadership and technical product direction at DeepMind. Beyond Google, he helps grow the ecosystem as an organizer of the M2L ML summer school and as an honorary lecturer teaching postgraduate reinforcement learning. His background in mathematical engineering and machine learning gives him a rare mix of theoretical depth and production-grade systems craftsmanship.
11 years of coding experience
11 years of employment as a software developer
Classical High School Diploma Humanities, Classical High School Diploma Humanities at Liceo Classico Cesare BeccariaMaster of Science (MSc) Computer Science Engineering, Master of Science (MSc) Computer Science Engineering at Politecnico di MilanoUniversity College London
Italian, English
Github Skills (25)
algorithm10
algorithms10
activation-function10
probabilistic-programming10
gradient-descent10
python10
optimizers10
testing10
machine-learning10
optimizer10
reinforcement-learning10
numpy10
deep-learning10
tensorflow10
optimisation10
Programming languages (2)
Github contributions (5)
google-deepmind/rlax
Jan 2020 - Nov 2022
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 35 commits, 4 PRs in 2 years 9 months
Contributions summary:Matteo contributed to the `rlax` repository by formatting and updating existing test files within the `src` directory. These updates primarily involve adjusting the layout and structure of test functions across various value learning and distribution tests, including TDLearning, TDLambda, Sarsa, and QLearning, alongside those for Retrace and L2Project. The changes involve formatting code and test cases. This indicates a focus on maintaining code quality and ensuring the correctness of reinforcement learning algorithms within the `rlax` framework.
google-deepmind/optax
Sep 2020 - Sep 2022
Optax is a gradient processing and optimization library for JAX.
Role in this project:ML Engineer
ML EngineerContributions:1 release, 110 reviews, 57 commits in 2 years
Contributions summary:Matteo primarily contributed to the Optax library by adding and modifying optimization algorithms, specifically focusing on gradient processing and optimization techniques for JAX. Their work included the implementation of AdaGrad, and various other optimization algorithms such as AdaBelief and LARS. They also addressed pytyping issues, added type definitions and corrected formatting. Furthermore, they removed deprecated code, and added utilities for eigenvector and matrix inverse pth root computation.
jaxmachine-learningoptimization
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.