Michael Oliver
Chief Scientist at Numerai
Woodinville, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolMichael Oliver is Chief Scientist at Numerai with 11 years of experience at the intersection of computational neuroscience and applied machine learning. Trained as a Ph.D. in Computational Neuroscience at UC Berkeley, he moved from recording and analyzing visual cortex electrophysiology to building production-grade ML systems and leading research teams. At Numerai he progressed rapidly from Data Scientist to Minister of Research and now leads scientific strategy, blending foundational research with market-facing modelling. He is an active contributor to the Keras ecosystem—improving optimizers like Adam and adding advanced activations and cosine-normalized layers—reflecting deep expertise in neural network internals and numerical robustness. Based in Woodinville, WA, he combines academic rigor with practical engineering: debugging memory leaks, writing tests, and refactoring core libraries to make cutting-edge ideas usable in production. Colleagues rely on him for translating complex theory into stable, reproducible code and scalable models.
11 years of coding experience
15 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Computational Neurosceince, Doctor of Philosophy (Ph.D.) Computational Neurosceince at University of California, Berkeley
Github Skills (16)
neural-network10
keras10
machine-learning10
optimizer10
constraint10
deep-learning10
tensorflow10
python10
artificial-neural-networks10
optimizers10
regularization10
theano9
deeplearning-ai9
data-science8
jax7
Programming languages (4)
Github contributions (5)
keras-team/keras-contrib
Jan 2017 - Aug 2017
Keras community contributions
Role in this project:
ML Engineer
ML EngineerContributions:69 commits, 61 PRs, 85 pushes in 6 months
Contributions summary:Michael primarily contributed to the `keras-contrib` repository by addressing issues and implementing functionalities related to advanced activation functions, specifically the PELU (Parametric Exponential Linear Unit) and SReLU (S-shaped Rectified Linear Unit). Their work involved debugging and refactoring code, as well as adding features such as a Clip constraint and tests. Furthermore, the user has added a cosine normalized dense layer and corrected cosine convolution.
pythondata-sciencedeep-learningtheanoneural-networks
keras-team/keras
Apr 2015 - Mar 2017
Deep Learning for humans
Role in this project:ML Engineer
ML EngineerContributions:37 commits, 29 PRs, 132 comments in 1 year 11 months
Contributions summary:Michael made significant contributions to the Keras library, demonstrating expertise in deep learning and optimization techniques. They focused on improving the Adam optimizer by fixing bugs and aligning it with the original paper. Furthermore, the user added support for regularization, constraints, and KL divergence, enhancing the library's flexibility and functionality for various machine learning tasks. They also addressed memory leaks and refactored the codebase for clarity.
pythondata-sciencedeep-learningneural-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.