Raphael Meudec
Research Scientist at Blue Morpho
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolRaphael Meudec is a research scientist and machine learning engineer with 11 years of experience bridging academic research and production-grade ML systems. He completed a PhD at Inria focused on automated meta-analysis of fMRI studies, and has contributed to major open-source projects such as Keras and Nilearn—improving core neural network behaviors, visualization tools, and test suites. At Blue Morpho he builds enterprise knowledge graphs that let LLMs navigate proprietary data beyond simple RAG, combining representation learning with practical engineering. His background spans optimization and data science from ENSTA ParisTech and Université Paris-Saclay, and he has a track record of implementing advanced models (GANs, recurrent layers) and robust training pipelines. Notably, he has acted as a Google Developer Expert in ML, reflecting both community impact and deep technical expertise.
11 years of coding experience
4 years of employment as a software developer
Master 2, Optimization, Operational Research and Commands, Applied Mathematics, Master 2, Optimization, Operational Research and Commands, Applied Mathematics at ENSTA ParisTech - École Nationale Supérieure de Techniques AvancéesMaster 2, Data Science, Master 2, Data Science at Université Paris-SaclayStanislas
English, Spanish, French
Stackoverflow
696reputation
54kreached
33answers
0questions
tensorflow
top-5%
Github Skills (28)
python10
dcgan10
brain-imaging10
testing10
scikit10
machine-learning10
cgan10
generative-adversarial-network10
reinforcement-learning10
keras10
deep-learning10
tensorflow10
neuroimaging10
scikit-learn10
neural-network10
Programming languages (6)
Github contributions (5)
RaphaelMeudec/deblur-gan
Feb 2018 - Dec 2019
Keras implementation of "DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks"
Role in this project:
ML Engineer
ML EngineerContributions:35 commits, 6 PRs, 25 pushes in 1 year 10 months
Contributions summary:Raphael contributed significantly to the Keras-based deblurring model. They implemented core features such as residual blocks and reflection padding layers, and refactored the training loop, including the integration of Wasserstein loss and gradient penalty loss, demonstrating a strong understanding of GAN architectures and optimization techniques. The user also added functionalities for logging and saving model weights, along with the addition of a script to process the GOPRO_Large dataset.
blinddeep-learningadversarialconditionaldeblurring
keras-rl/keras-rl
Mar 2018 - Jan 2019
Deep Reinforcement Learning for Keras.
Role in this project:ML Engineer
ML EngineerContributions:2 releases, 42 commits, 34 PRs in 10 months
Contributions summary:Raphael made several updates and modifications related to the Keras-RL library. The commits updated links to the source code, adjusted code to be compatible with different versions of Keras and Gym, and reorganized directories. Further contributions included adding documentation for the rl.core methods, and memory and policy modules, as well as fixing an issue with the tensorboard callback.
deep-learningreinforcement-learningtheanoneural-networksmachine-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.