Raphaël Marinier
Staff Software Engineer at Google
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolRaphaël Marinier is a Staff Software Engineer based in Paris with a decade of experience building large-scale, production-grade systems and research-driven ML infrastructure at Google. He has led teams across Google Brain, YouTube, and Maps, shipping real-time traffic modeling, scalable search and channels infrastructure, and advances in reinforcement learning and audio generation. Pragmatic in both research and engineering, he contributes to high-profile open-source projects like DeepMind’s OpenSpiel and SEED RL, improving Python/C++ interoperability, MFG support, and performance profiling for accelerated RL. His background in performance optimization also spans desktop software—reducing CPU usage in Audacity through targeted rendering changes—showing a knack for system-level efficiency gains. Trained at École Polytechnique, he combines strong theoretical foundations with hands-on optimizations and a track record of moving research prototypes into robust, production systems.
9 years of coding experience
9 years of employment as a software developer
Master of Science - MS, Computer Science, Master of Science - MS, Computer Science at École Polytechnique
French, English, Spanish
Github Skills (25)
fftw10
c-language10
python10
ata10
coregames10
machine-learning10
videogames10
wxwidgets10
fft10
tensorflow10
performance-optimization10
atari260010
cprogramming-language10
api-design9
reinforcement-learning9
Programming languages (7)
Github contributions (5)
google-research/seed_rl
Oct 2019 - Nov 2020
SEED RL: Scalable and Efficient Deep-RL with Accelerated Central Inference. Implements IMPALA and R2D2 algorithms in TF2 with SEED's architecture.
Role in this project:
ML Engineer
ML EngineerContributions:22 commits, 1 PR, 13 pushes in 1 year 1 month
Contributions summary:Raphaël primarily contributed to the core infrastructure of the SEED RL project. They implemented profiling utilities for performance analysis within the actor tasks, demonstrating an understanding of the project's architecture. The user optimized Atari environment interactions using bit-packing techniques and frame stacking, demonstrating knowledge of efficient data handling. They also made changes to support the merging of actor and learner files across environments, which indicates a focus on streamlining the codebase and potentially improving the training workflow.
gcpstable-baselinesdeep-reinforcement-learningdeepmind-labdeep-rl
google-deepmind/open_spiel
May 2021 - Aug 2021
OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.
Role in this project:Back-end Developer
Back-end DeveloperContributions:28 reviews, 22 commits, 42 comments in 3 months
Contributions summary:Raphaël primarily focused on enhancing the functionality of the Python components within the OpenSpiel project, specifically concerning the behavior of observation strings and related functions to align them with the C++ environments. They also made improvements to the code for the Mean Field Games (MFG) within the project, addressing comments and API compliance. The user contributed towards adding support for MFGs, including enabling them for playthrough and state sampling, and improving the tests.
cppmultiagentgamespythondatamining
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.