Kristian Holsheimer
Research Engineer at Google DeepMind
London, England, United Kingdom
Summary
👤
Senior🎓
Top SchoolKristian Holsheimer is a research engineer at DeepMind with 11 years of experience applying rigorous mathematical thinking to real-world machine learning systems. He holds a PhD in string theory and has transitioned that theoretical rigor into pragmatic, production-focused roles across Microsoft, Booking.com and startups, building everything from fraud and ad-bidding models to deployed malware-detection systems. Kristian contributes to notable open-source projects such as DeepMind’s dm-haiku, improving core JAX neural network components and ensuring efficient, testable implementations of modules like MultiHeadAttention. He enjoys bridging abstract mathematics and applied ML—often reaching for pen and paper to work out proofs or discussing subtleties with peers—and maintains keras-gym to make reinforcement learning more accessible. Based in London, he combines deep academic expertise with hands-on engineering to move models from concept to impact.
11 years of coding experience
10 years of employment as a software developer
Bachelor of Science (BSc), Physics and Astronomy, Bachelor of Science (BSc), Physics and Astronomy at University of AmsterdamMSc Research Exchange, Theoretical Physics, MSc Research Exchange, Theoretical Physics at McGill University
English, Dutch
Stackoverflow
22,779reputation
8.4mreached
69answers
15questions
tensorflow
top-5%
pandas
top-1%
wxpython
top-1%
matplotlib
top-5%
python
top-1%
permutation
top-1%
Github Skills (25)
permutation10
python10
pandas10
machine-learning10
dataframe10
deep-learning10
shuffle10
wxpython10
neural-network10
jax10
matplotlib9
seaborn9
tensorflow9
immutability-helper9
immutability9
Programming languages (7)
Github contributions (5)
google-deepmind/dm-haiku
Mar 2020 - Mar 2021
JAX-based neural network library
Role in this project:
ML Engineer
ML EngineerContributions:6 commits, 5 PRs, 11 comments in 1 year
Contributions summary:Kristian contributed to the `dm-haiku` library by implementing and improving core functionalities related to neural networks. Their work includes refactoring code to enhance immutability with `frozendict` and `__slots__`. They also addressed edge cases in `Transformed` and `MultiHeadAttention` modules, ensuring compatibility and efficient use within the JAX-based deep learning framework. Further, they added tests demonstrating the vmap functionality within MultiHeadAttention.
deep-learningneural-networksmachine-learningneural-networkdeep-neural-networks
KristianHolsheimer/keras-gym
Mar 2019 - Sep 2020
Plug-n-play reinforcement learning with OpenAI Gym and Keras
Contributions:302 commits, 9 PRs, 188 pushes in 1 year 7 months
deep-learningreinforcement-learningplug-n-playdeep-reinforcement-learningopenai
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.