Kai Arulkumaran

Research Scientist at Sakana AI

Chiyoda, Japan
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

🤩
Rockstar
🎓
Top School
Kai Arulkumaran is a research scientist with 13 years of experience bridging deep reinforcement learning research and production-ready tooling, currently based in Chiyoda, Japan. With a PhD in Bioengineering from Imperial College London and a BA in Computer Science from Cambridge, he combines rigorous academic training with hands-on ML engineering—contributions to notable projects include implementing categorical DQN, dueling and noisy layers in the well-known Rainbow RL codebase and improving PyTorch RL and VAE examples. He has led research teams at Araya and held internships at DeepMind, Facebook, and Microsoft, demonstrating a track record of moving ideas from prototype to scalable experiments. Equally fluent in DevOps, his dockerfiles repository shows careful attention to CUDA environments, remote access, and reproducible ML stacks. Outside core research he identifies as a programmer, DJ, and transhumanist, reflecting a curiosity-driven mindset that informs unconventional problem solving.
code13 years of coding experience
job13 years of employment as a software developer
bookDoctor of Philosophy (Ph.D.), Bioengineering, Doctor of Philosophy (Ph.D.), Bioengineering at Imperial College London
bookBachelor of Arts (BA), Computer Science, Bachelor of Arts (BA), Computer Science at University of Cambridge
github-logo-circle

Github Skills (17)

pytorch10
docker10
vnc10
variational-autoencoder10
machine-learning10
bash10
dockers10
reinforcement-learning10
autoencoder10
dqn10
deep-learning10
cuda10
deep-reinforcement-learning10
ssh10
linux9

Programming languages (16)

JavaC++CSSCCMakeGoHTMLJupyter Notebook

Github contributions (5)

github-logo-circle
Kaixhin/dockerfiles

Feb 2015 - Aug 2019

Compilation of Dockerfiles with automated builds enabled on the Docker Registry
Role in this project:
userDevOps Engineer
Contributions:362 commits, 6 PRs, 342 pushes in 4 years 5 months
Contributions summary:Kai primarily focused on creating and maintaining Dockerfiles for various CUDA versions and related tools. Their contributions include setting up VNC servers for remote access, configuring SSH access, and integrating ceph scripts. They also implemented environment variable configurations for CUDA libraries and addressed issues related to VNC restarts.
compilationbuildsdockerhydrogenautomated-builds
Kaixhin/Rainbow

Oct 2017 - Nov 2020

Rainbow: Combining Improvements in Deep Reinforcement Learning
Role in this project:
userML Engineer
Contributions:6 releases, 4 reviews, 145 commits in 3 years 1 month
Contributions summary:Kai primarily contributed to the deep reinforcement learning project by implementing core functionality and integrating new features. Their work involved updating the project's requirements, adding an environment wrapper and implementing a categorical DQN. Furthermore, they addressed the architecture, by adding dueling and noisy layers to the model. These changes reflect an iterative approach to improving the deep reinforcement learning model.
deep-learningreinforcement-learningrainbowdeep-reinforcement-learningimprovements
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.
Request Free Trial
Kai Arulkumaran - Research Scientist at Sakana AI