Léonard Berrada
Staff Research Scientist at Google DeepMind
England, United Kingdom
Summary
👤
Senior🎓
Top SchoolLéonard Berrada is a Staff Research Scientist at Google DeepMind with 11 years of experience applying theory-driven methods to large-scale deep learning and optimization. His work spans LLM pre-training (currently on Gemini-scale pre-training), state-space models like Griffin / Recurrent Gemma, and practical privacy-preserving training—co-creating the jax_privacy library. Trained with a PhD supervised by Andrew Zisserman and Pawan Kumar, he brings rigorous optimization and formal verification expertise to production-scale model training. He has a track record of shipping research code and tools that bridge theory and systems, from recurrent SSMs to differential privacy at scale. Based in England, he blends strong academic foundations (Oxford, Berkeley, CentraleSupélec) with hands-on engineering at DeepMind, and is known for making mathematically grounded ideas practically deployable.
11 years of coding experience
6 years of employment as a software developer
Master of Engineering (M.Eng.) Industrial Engineering and Operations Research, Master of Engineering (M.Eng.) Industrial Engineering and Operations Research at University of California, BerkeleyMaster of Science (M.S.) Bachelor of Science (B.S) Engineering, Master of Science (M.S.) Bachelor of Science (B.S) Engineering at CentraleSupélecAdvanced Mathematics & Physics Mathematics - Physics - Engineering science, Advanced Mathematics & Physics Mathematics - Physics - Engineering science at Lycée Sainte GenevièveDoctor of Philosophy (Ph.D.) Autonomous Intelligent Machines and Systems, Doctor of Philosophy (Ph.D.) Autonomous Intelligent Machines and Systems at University of OxfordBachelor of Science (BS) Physics, Bachelor of Science (BS) Physics at Paris-Sud University (Paris XI)
French, English, Spanish
Github Skills (47)
loss-functions10
python10
machine-learning10
numpy10
tensorboard10
deep-learning10
gpu10
autograd10
neural-network10
logging10
tensor10
jax10
gpu-acceleration10
visdom10
crayon9
Programming languages (5)
Github contributions (5)
oval-group/smooth-topk
Jan 2019 - Aug 2021
Smooth Loss Functions for Deep Top-k Classification
Contributions:2 releases, 7 commits, 5 PRs in 2 years 7 months
classificationloss-functions
oval-group/mlogger
Mar 2017 - Nov 2020
a lightweight and simple logger for Machine Learning
Contributions:3 releases, 1 review, 10 commits in 3 years 8 months
machine-learningloggingpythonvisdomdeep-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.