Kate Bäumli
Staff Research Engineer at Google DeepMind
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolKate Bäumli is a Staff Research Engineer at Google DeepMind with a decade of experience building reinforcement learning systems and agent architectures. Trained in Electrical and Computer Engineering at UT Austin, she joined DeepMind as a research engineer in 2019 and has advanced through roles focused on scalable, JAX-based RL implementations. Her open-source contributions to flagship DeepMind libraries like acme and rlax show hands-on improvements—from a JAX R2D2 agent and Atari network adaptations to performance optimizations such as static memory allocation and distrax migration. Based in London, she blends research rigor with production-minded engineering to help agents acquire and efficiently reuse knowledge, and she often surfaces subtle but impactful fixes (shape flexibility, testing, and doc corrections) that improve long-running experiments.
9 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree, Electrical and Computer Engineering, Bachelor’s Degree, Electrical and Computer Engineering at The University of Texas at Austin
Github Skills (11)
neural-network10
machine-learning10
agent10
deep-learning10
jax10
python10
distributions10
reinforcement-learning10
optimization10
testing9
tensorflow5
Programming languages (4)
Github contributions (5)
google-deepmind/rlax
Oct 2020 - Jan 2023
Role in this project:
ML Engineer
ML EngineerContributions:7 releases, 2 reviews, 39 commits in 2 years 3 months
Contributions summary:Kate focused on improving the efficiency and functionality of the rlax library. Their contributions include optimizing intrinsic reward calculations, specifically by switching to static memory allocation. They also refactored and enhanced V-Trace and MPO functions for reinforcement learning, introducing shape flexibility and fixing docstrings. Furthermore, the user contributed significantly to the distributions module, migrating it to use distrax and adding tests, including tests for squashed gaussian distributions.
google-deepmind/acme
Jul 2020 - Mar 2021
A library of reinforcement learning components and agents
Role in this project:ML Engineer
ML EngineerContributions:7 commits in 7 months
Contributions summary:Kate contributed significantly to the `acme` repository, which focuses on reinforcement learning components. Their work included adding a JAX implementation of the R2D2 agent, modifying Atari network architectures and utilities to work with JAX, and making changes to the actor implementation to handle batched and unbatched inputs. Further, they updated the distributional networks package, including the addition of new heads. These changes indicate a focus on expanding the library's capabilities and improving its flexibility for RL experiments.
reinforcement-learningreinforcementagentsdeep-reinforcement-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.