Jonathan Godwin
CEO at Orbital
England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolJonathan Godwin is a machine learning leader and founder with nine years of experience bridging cutting-edge research and product-focused engineering, currently serving as CEO of Orbital. Previously a researcher at DeepMind, he co-authored high-profile papers on mammography and physics simulation and created jraph, a widely used graph neural network library in Jax that powers reproducible graph ML work. His ICLR 2022 "noisy nodes" paper advanced state-of-the-art on quantum chemistry tasks and was adopted in major competitions, reflecting a blend of theoretical insight and practical impact. With an MSc in Machine Learning from UCL and an early career in NLP and social finance, he pairs deep technical expertise with entrepreneurial instincts and a knack for turning complex simulations into usable tools.
9 years of coding experience
9 years of employment as a software developer
Mathematics & Philosophy, Mathematics & Philosophy at University of BristolAlleyn'sUniversity College London
Github Skills (6)
machine-learning10
jax10
graph-neural-network10
python10
flax9
deep-learning9
Programming languages (2)
Github contributions (5)
google-deepmind/jraph
Dec 2020 - Aug 2022
A Graph Neural Network Library in Jax
Role in this project:
ML Engineer
ML EngineerContributions:11 releases, 14 reviews, 50 commits in 1 year 9 months
Contributions summary:Jonathan primarily focused on setting up documentation for the Jraph library, utilizing Sphinx. They contributed to utility functions within the library, including batching, unbatching, padding, and masking functionalities, indicating a focus on graph data handling. Additionally, the user introduced a utility function for generating fully connected graphs, and integrated a flax version of an OGB molhiv example, showcasing expertise in machine learning and the Jraph library. Further commits added a zero out padding value function, which suggests an effort towards improving the numerical stability of the model.
deep-learningneural-graphgraph-neural-networkmachine-learninggraph-neural-networks
jg8610/multi-task-learning
Jul 2016 - Dec 2017
Contributions:12 commits, 10 pushes, 1 branch in 1 year 5 months
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.