Andrew Brock
Research Scientist at DeepMind
United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolAndrew Brock is a Research Scientist at DeepMind with a decade of experience training and stabilizing large-scale neural networks, notable for significant contributions to the widely referenced BigGAN-PyTorch implementation. He blends deep ML engineering—fixing EMA and batchnorm instabilities and adding launch configurations for ImageNet-scale experiments—with a strong systems and control background from mechatronics and embedded haptics work. His GitHub handle, “Dimensionality Diabolist,” hints at a focus on high-dimensional generative modeling and latent-space manipulation, exemplified by contributions to a neural photo editor and novel image-manipulation utilities. Trained as an MS in Mechanical Engineering, he pairs theoretical rigor with practical simulation and control skills developed at HaptX/Axon VR and in academic teaching roles. Outside ML, he has long practiced instructive roles—from ballroom tango instructor to university teaching assistant—bringing clear communication and pedagogy to complex technical problems.
10 years of coding experience
3 years of employment as a software developer
California Polytechnic State University, San Luis Obispo
Github Skills (15)
neural-network10
computer-vision10
pytorch10
machine-learning10
deep-learning10
cgan10
cyclegan10
python10
image-processing10
theano10
dcgan10
lasagne10
apidoc9
api9
tensorflow8
Programming languages (3)
Github contributions (5)
ajbrock/BigGAN-PyTorch
Jan 2019 - Jul 2019
The author's officially unofficial PyTorch BigGAN implementation.
Role in this project:
ML Engineer
ML EngineerContributions:83 commits, 6 PRs, 30 pushes in 5 months
Contributions summary:Andrew made several significant contributions to the BigGAN-PyTorch repository, including the initial upload of core files and updates to utility functions and inception moment calculations. They also fixed bugs related to Exponential Moving Average (EMA) and batch normalization, crucial for stable training. Moreover, the user added launch scripts for the SNGAN model and various ImageNet configurations, expanding the repository's functionality.
pytorchdeep-learningbigganneural-networksgans
ajbrock/Neural-Photo-Editor
Sep 2016 - Mar 2017
A simple interface for editing natural photos with generative neural networks.
Role in this project:ML Engineer
ML EngineerContributions:22 commits, 4 PRs, 20 pushes in 6 months
Contributions summary:Andrew primarily contributed to the development of a neural photo editor. Their work involved creating and integrating theano functions for image manipulation, including functions for latent space manipulation and color gradient calculations. The user also added support for different model architectures (IAN), including integrating a plat interface for framework independence, and implemented subpixel layers. They also optimized the code by replacing cuDNN layers with equivalent transposed convolutional layers.
pythoninpaintingneural-networksimage-processinggenerative
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.