David Warde-farley
Staff Research Scientist at Google DeepMind
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolDavid Warde-Farley is a Staff Research Scientist at Google DeepMind based in London with 16 years of experience building and shipping machine learning research and infrastructure. He specializes in neural network research and production ML systems, focusing on algorithmic improvements, performance-critical C/Python optimizations, and robust tooling. An active open-source contributor to core scientific Python projects—including NumPy (GIL handling improvements and a minlength addition to bincount), Theano/PyTensor, Jupyter/IPython (implementing an "undo" for cell deletion), SciPy and scikit-learn—he consistently bridges research with developer-facing engineering. He progressed through Research Scientist roles at DeepMind after internships on Google Brain and Google Photos, pairing applied research with scalable implementations. He holds a PhD in Computer Science from Université de Montréal and combines rigorous academic training with practical work that makes ML algorithms more reliable and reproducible.
17 years of coding experience
6 years of employment as a software developer
PhD, Computer Science, PhD, Computer Science at Université de MontréalHon. B. Sc., Computer Science, Hon. B. Sc., Computer Science at University of Toronto
Github Skills (58)
algorithm10
unit-testing10
algorithms10
notebook10
javascript10
restructuredtext10
data-pipelines10
python10
jupyter10
data-science10
sphinx10
scikit10
technical-writing10
testing10
machine-learning10
Programming languages (13)
Github contributions (5)
lisa-lab/pylearn2
Jul 2010 - Jun 2015
Warning: This project does not have any current developer. See bellow.
Role in this project:
ML Engineer
ML EngineerContributions:991 commits, 31 PRs, 10 pushes in 5 years
Contributions summary:David made several contributions related to the implementation of Restricted Boltzmann Machines (RBMs) and autoencoders within the pylearn2 framework. The commits include code for constructing RBMs with Gaussian-binary visible units, adding support for the identity activation function, implementing the stochastic gradient descent algorithm, defining and improving the performance of training for those networks, and other model building and testing functionalities. The user's work also involved the refactoring and enhancement of various modules, including those for building and using Deep Belief Networks.
javascripttypescript
mila-iqia/blocks
Feb 2015 - Nov 2016
A Theano framework for building and training neural networks
Role in this project:Back-end Developer
Back-end DeveloperContributions:306 commits, 174 PRs, 266 pushes in 1 year 9 months
Contributions summary:David primarily contributed to the Theano framework for building and training neural networks, focusing on improvements to the SimpleExtension class, including the addition of features like `every_n_epochs` and the handling of invalid condition keywords. They also addressed flake8 errors, refactored code, and made modifications to testing files by setting up the right conditions for model evaluation and serialization of roles. Furthermore, they added support for Theano known_grads and parameter clipping, improving the functionality of the GradientDescent class and algorithm performance.
pytorchdeep-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.