Yuta Kikuchi - Researcher at Preferred Networks, Inc.

Yuta Kikuchi

Researcher at Preferred Networks, Inc.

Tokyo, Japan

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Yuta Kikuchi is a researcher and machine learning engineer based in Tokyo with 10 years of experience applying deep learning and numerical computing to real-world problems. Holding a PhD in Natural Language Processing from Tokyo Institute of Technology, he transitioned from academic research to industry at Preferred Networks, where he focuses on robust, high-performance ML systems. He is an active open-source contributor to prominent projects such as CuPy and Chainer/ChainerRL, improving GPU-backed array operations, matrix algebra correctness, and reinforcement learning data pipelines. His work blends low-level array manipulation fixes and comprehensive test coverage with higher-level algorithmic refactors, showing both attention to numerical correctness and maintainability. Colleagues rely on him for hard-to-reproduce bug fixes in linear algebra on GPUs and for making RL agents’ data handling more efficient. He brings a rare combination of NLP research training and hands-on systems engineering in GPU-accelerated ML stacks.

10 years of coding experience

Doctor of Philosophy (PhD), Natural Language Processing, Doctor of Philosophy (PhD), Natural Language Processing at Tokyo Institute of Technology

Bachelor of Engineering (BEng), cognitive, Bachelor of Engineering (BEng), cognitive at Kisarazu National College of Technology

Japanese, English

Github Skills (17)

unit-testing10

python10

chainer10

testing10

data-manipulation10

machine-learning10

reinforcement-learning10

numpy10

deeplearning-ai10

deep-learning10

gpu10

cupy10

cuda10

linear-algebra10

dqn9

Programming languages (2)

C#Python

Github contributions (5)

chainer/chainer

Nov 2015 - Mar 2016

A flexible framework of neural networks for deep learning

Role in this project:

ML Engineer

Contributions:11 commits, 5 PRs, 4 pushes in 3 months

Contributions summary:Yuta primarily focused on enhancing the functionality and stability of the `chainer/chainer` deep learning framework. Their contributions included fixing transpose functions, addressing bugs in matrix multiplication operations, and improving the handling of `cupy.ndarray`. The user also implemented new tests to ensure the correct behavior of these functions, specifically incorporating tests for broadcasted matrix operations.

cudapythonmxnetcaffe2flexible-framework

chainer/chainerrl

Feb 2017 - Feb 2017

ChainerRL is a deep reinforcement learning library built on top of Chainer.

Role in this project:

ML Engineer

Contributions:22 commits, 3 PRs, 2 branches in 18 days

Contributions summary:Yuta primarily focused on refactoring and modifying core components of the ChainerRL library, specifically within the DQN and DDPG agent implementations. Their contributions included separating and modifying the `batch_states` method for improved data handling and code organization, demonstrating a focus on performance and maintainability within the reinforcement learning framework. These changes touch upon core aspects of data preparation within the reinforcement learning agents. The user updated various agents by adding and improving the implementation of `batch_states`.

actor-criticdeep-learningreinforcement-learningdeep-reinforcement-learningchainer

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