Yujin Tang
Research Scientist at Sakana AI
Tokyo, Japan
Summary
🤩
Rockstar🎓
Top SchoolYujin Tang is a research scientist in Tokyo with nine years of experience advancing reinforcement learning and robotics, currently applying RL research to real-world robotics problems at Sakana AI. Previously embedded in Google and DeepMind research teams, Yujin has deep expertise in designing and tuning actor-critic architectures (DDPG/TD3) and building comprehensive agent frameworks for robust experimentation. Contributions to high-profile open-source projects such as GoogleCloudPlatform/cloudml-samples and EvoJAX demonstrate both practical engineering—fixing distributed pmap issues and integrating Brax examples—and a knack for reproducible experimentation. Yujin holds advanced degrees from The University of Tokyo and Waseda University, pairing rigorous academic training with production-grade ML engineering. Colleagues describe Yujin as someone who bridges custom neural architecture work with engineering pragmatism, often surfacing subtle bugs in distributed simulators that improve scalability and reliability.
9 years of coding experience
13 years of employment as a software developer
University of TokyoMaster's degree, Information Technology, Master's degree, Information Technology at Waseda UniversityBachelor's degree, Computer Science, Bachelor's degree, Computer Science at Shanghai Jiao Tong University
English, Chinese, Japanese
Github Skills (18)
pytorch10
python10
machine-learning10
reinforcement-learning10
ml10
mle10
tensorflow10
tdd10
ddpg10
neural-network10
jax10
gcp9
keras8
ata7
openai-gym7
Programming languages (4)
Github contributions (5)
google/evojax
Feb 2022 - Dec 2022
Role in this project:
ML Engineer
ML EngineerContributions:3 releases, 4 reviews, 56 commits in 10 months
Contributions summary:Yujin primarily contributed to the EvoJAX project by implementing and modifying components related to machine learning and reinforcement learning. Their work includes fixing an issue with `pmap` in the `SimManager` class, bumping version numbers, and adding a Brax tasks example and an MDKP example. They also changed the `PolicyNetwork` interface, implemented an example using MAP-Elites for a Brax Ant environment, and added example notebooks covering the implementation of algorithms, policy networks, and tasks.
google/brain-tokyo-workshop
Mar 2020 - May 2022
🧠🗼
Role in this project:ML Engineer
ML EngineerContributions:24 commits, 1 PR, 21 pushes in 2 years 1 month
Contributions summary:Yujin primarily contributes to developing and modifying machine learning models, particularly within the context of reinforcement learning. The user implements custom neural network layers and modifies existing environments for training and experimentation. The commits involve defining model architectures, implementing custom loss functions, and integrating these components within a training pipeline for reinforcement learning agents. The user's work includes experiment reproducibility and debugging.
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.