Geoffrey Négiar
Co-Founder at The Forecasting Company
Berkeley, California, United States
Summary
🤩
Rockstar🎓
Top SchoolGeoffrey Négiar is a co-founder and machine learning engineer based in Berkeley with 11 years of experience bridging deep research and product-focused ML systems. He holds a PhD in EECS from UC Berkeley and advanced mathematics and ML training from École Polytechnique and ENS Cachan, grounding him in rigorous theory and practical model-building. As co-founder of The Forecasting Company and a contributor to the prominent google/jaxopt project, he focuses on making robust, hardware-accelerated training workflows reliable and testable. His open-source work includes fixing subtle shape bugs and adding non-regression tests to strengthen production-ready examples, signaling attention to long-term maintainability. Comfortable moving between research, code, and startup execution, he brings a rare combination of academic depth and hands-on engineering to forecasting and robust ML tooling.
11 years of coding experience
Doctor of Philosophy - PhD, Electrical Engineering and Computer Science, Doctor of Philosophy - PhD, Electrical Engineering and Computer Science at University of California, BerkeleyMaster of Science (M.S.), Machine Learning, Computer Vision, Master of Science (M.S.), Machine Learning, Computer Vision at École Normale Supérieure de CachanManagement, Criminal Investigation, Management, Criminal Investigation at Military Police Officers Academy (EOGN)Baccalauréat S - High scool degree, Maths & Physics, Baccalauréat S - High scool degree, Maths & Physics at Lycée CondorcetMaster's degree, Mathematics and Computer Science, Master's degree, Mathematics and Computer Science at Ecole polytechniqueY CombinatorHigher School Preparatory Classes, Mathematics, Physique, Higher School Preparatory Classes, Mathematics, Physique at Lycée Louis le Grand
French, English, Spanish, Russian, Japanese
Stackoverflow
809reputation
113kreached
16answers
14questions
Github Skills (14)
differentiable-programming10
deep-learning10
jax10
optimization10
python9
numpy8
tensorflow-datasets7
flax7
machine-learning7
cvxpy6
pytorch6
torchtext6
nlp6
dataframe6
Programming languages (10)
Github contributions (5)
google/jaxopt
Nov 2021 - Jan 2022
Hardware accelerated, batchable and differentiable optimizers in JAX.
Role in this project:
ML Engineer
ML EngineerContributions:16 reviews, 6 commits, 7 PRs in 2 months
Contributions summary:Geoffrey primarily focused on modifying and improving an example for robust deep learning training within the JAXopt library. Their contributions included updating the epoch logic for training, incorporating feedback from another developer, and fixing a shape bug within the robust training example. Furthermore, they added a non-regression test, likely to prevent future issues with the same code. These changes demonstrate a focus on improving and validating the robustness of the provided deep learning examples.
pytorchdifferentiabledifferentiable-programmingautomatic-differentiationdeep-learning
openopt/chop
Nov 2020 - Nov 2021
CHOP: An optimization library based on PyTorch, with applications to adversarial examples and structured neural network training.
Contributions:1 release, 9 reviews, 311 commits in 1 year
pytorchadversarial-attackspythonstructureddeep-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.