Joseph Defretin
Data Scientist
Grenoble, Auvergne-Rhône-Alpes, France
Summary
🤩
Rockstar🎓
Top SchoolJoseph Defretin is a Data Scientist with 13 years of experience specializing in computer vision, image processing and deep learning, currently applying state-of-the-art models at Probayes to deliver practical visual perception solutions across industries. He holds a PhD in applied mathematics and has a strong research-to-production trajectory—developing active vision frameworks in academia and deploying forecasting and scene-understanding systems in industry. His open-source contributions span high-profile scientific Python projects (numpy, spyder) and GPU-enabled tooling, where he’s improved compatibility, testing and demos to help others run GPU-accelerated workflows. Comfortable across full-stack and backend work, he combines rigorous mathematical foundations with pragmatic engineering for robust model pipelines and real-world deployment. An often-overlooked strength is his history of improving tooling and documentation, which amplifies team productivity and long-term maintainability.
13 years of coding experience
6 years of employment as a software developer
Electrical engineering, Electrical engineering at École Supérieure d'ÉlectricitéClasses préparatoires maths sup / maths spé, Classes préparatoires maths sup / maths spé at Lycée César Baggio, LillePhD, Applied mathematics, PhD, Applied mathematics at École Normale Supérieure de Cachan
English, French
Github Skills (27)
debugging10
debug10
platform-agnostic10
develop10
qt10
multiplatform10
python10
pycuda10
sphinx10
testing10
pyside10
ide10
cross-platform10
numpy10
regression-testing10
Programming languages (12)
Github contributions (5)
spyder-ide/spyder
Feb 2015 - May 2022
Official repository for Spyder - The Scientific Python Development Environment
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:23 PRs, 3 pushes, 957 comments in 7 years 3 months
Contributions summary:Joseph primarily contributed to the Spyder IDE's core functionality. They addressed exceptions within the `bsdsocket.read_packet` function and added a warning to the external console when an incompatible Python interpreter is selected. Furthermore, they updated the `path.py` library to version 7.3 and made several improvements related to the GitHub issue template. Finally, they worked on refactoring the plugin loading and improved the handling of different Python versions.
pythonpyqt5anaconda-distributionanacondapyqt
numpy/numpy
Oct 2013 - Jun 2019
The fundamental package for scientific computing with Python.
Role in this project:QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:11 commits, 1 PR, 40 comments in 5 years 9 months
Contributions summary:Joseph primarily contributed to improving the quality and reliability of the NumPy library through test-related commits. They fixed bugs related to specific functionalities like complex number handling in `cov/corrcoef` and old-style class compatibility in Python 3. The user also added regression tests to ensure the stability of bug fixes.
lapackpythonmpindarrayconvolution
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.