Sander Dieleman
Research Scientist (Director) at Google DeepMind
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolSander Dieleman is a research scientist and director at Google DeepMind with 17 years of experience applying deep learning to challenging problems, particularly in music information retrieval and recommender systems. His career spans academic research (PhD work on feature learning for MIR), industry internships at Spotify, and influential open-source contributions to core ML tooling such as Theano and Lasagne—where he implemented critical gradient and convolutional components and improved cuDNN and sparse support. Based in London, he combines rigorous research instincts with hands-on engineering, shipping production-facing models and low-level library optimizations. Beyond academia and industry, he founded and ran a music community portal for nearly a decade, reflecting a sustained practical interest in music technology that informs his research.
17 years of coding experience
6 years of employment as a software developer
Master Engineering Computer Science ICT, Master Engineering Computer Science ICT at Ghent University
Dutch, English, French, Portuguese, Swedish
Github Skills (17)
convolutional-neural-networks10
operation10
python10
tensorrt10
machine-learning10
deep-learning10
tensorflow10
neural-network10
tensor10
theano10
testing9
data-structure9
algorithm9
algorithms9
optimization9
Programming languages (3)
Github contributions (5)
Lasagne/Lasagne
Sep 2014 - Nov 2016
Lightweight library to build and train neural networks in Theano
Role in this project:
Back-end Developer
Back-end DeveloperContributions:276 commits, 144 PRs, 209 pushes in 2 years 2 months
Contributions summary:Sander's contributions primarily focused on building and expanding the functionality of a deep learning library. They implemented and refined core components such as layers, non-linearities, and update functions. The user also added functionality for specific layer types, including 1D and 2D convolutions, and contributed to the supporting infrastructure such as parameter initialization strategies and utility functions.
pythondeep-learningtheanoneural-networksdeep-learning-library
Theano/Theano
Feb 2014 - Mar 2016
Theano was a Python library that allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently. It is being continued as PyTensor: www.github.com/pymc-devs/pytensor
Role in this project:ML Engineer
ML EngineerContributions:5 commits, 24 comments in 2 years
Contributions summary:Sander contributed significantly to the Theano library, focusing on the implementation and optimization of gradient-related functionalities. They introduced the `consider_constant` operation, which truncates gradients, and implemented optimizations to remove it from the computation graph. The user also wrote tests to verify the correct gradient computations and the removal of the op. Furthermore, they made modifications to accommodate sparse variables and enhance compatibility with cuDNN.
python-librarymathmulti-dimensionalpythonevaluate
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.