Kento Aoyama

IT Infrastructure Engineer

Work Experience and Education

IT Infrastructure Engineer

CyberAgent, Inc.
2022 - Present

Development and operation of the ML Platform, a private machine learning platform on private HPC infrastructure, which provides various MLOps services such as distributed job, inference instance, and Jupyter environments for ML/DS Engineers.

Achievements:

  • Private Cloud Meetup #4 (Tokyo, Japan, 2023)
  • Cloud Operator Days Tokyo 2024 (Tokyo, Japan, 2024)

Technology keywords:

  • ML Platform Engineering
  • Kubernetes
  • golang
  • grpc
  • GPU/CUDA
  • MPI/NCCL
  • GPU Cluster Management

Researcher

School of Computing, Tokyo Institute of Technology
2021 - 2022

Ph.D. student

Department of Computer Science, School of Computing, Tokyo Institute of Technology
2016 - 2021

Research & development of Linux containers for High-Performance Computing applications, and implementation of the parallel/distributed system for bioinformatics software on clouds and supercomputers. (e.g. TSUBAME, ABCI, MS Azure)

Achievements:

The research results were presented at academic journals/conferences/tech-reports.

  • Supercomputing Frontiers and Innovations (2020)
  • Supercomputing Asia 2020 (Singapore, Feb. 2020)
  • Technical Program in SC19: The International Conference for High Performance Computing, Networking, Storage, and Analysis (Denver, CO, US, Nov. 2019)
  • The 25th International Conference on Parallel and Distributed Processing Techniques and Application (Las Vegas, NV, US, Jul. 2019)

Technology keywords:

  • Linux containers
  • Docker
  • Singularity
  • Shifter/Sarus
  • OCI container
  • OCI image
  • OCI registry
  • MPI/OpenMP
  • CUDA
  • Cloud Computing
  • Supercomputers
  • Bioinformatics
  • HPC Engineering

Research Internship

Swiss National Supercomputing Centre (CSCS), ETH Zurich
2017.9 - 2017.12

Research collaboration with HPC System Integration Group at CSCS. Designing and developing the software architecture of "Shifter-NG" (Next Generation), an open source Linux container for HPC environments. The internal architectural designs were inherited to the new OCI compatible container "Sarus".

Achievements:

The research results are used in the production environment of HPC system in CSCS.

  • HPC System at Swiss National Supercomputing Centre.
  • HPC-AI Advisory Council 2018, 2019.

Technology keywords:

  • Python
  • C/C++
  • Docker
  • Shifter/Sarus
  • OCI image
  • OCI registry

Software Developper

Fujitsu Ltd.
2015 - 2016

R&D for a cloud-integrated application platform service. Development of an Android version of the mobile platform which manages multi-media contents and web-based applications interacting with BaaS (Backend-as-a-Service).

Technology keywords:

  • Java
  • C/C++
  • Android
  • CI/CD
  • Agile Programming
  • DevOps Engineering

Master of Engineering

Department of Computer Science, Graduate School of Information Science and Engineering,
Tokyo Institute of Technology
2013 - 2015

Development of a Pipeline for Genome/Exome Analysis on the K computer. The pipeline can perform on over thousands of computing-nodes on the K computer and provides massively parallelized genome/exome sequencing analysis.

Achievements:

The research results were presented at academic journal and tech-report.

  • IPSJ Transactions on Advanced Computing System (ACS), 2016.
  • IPSJ SIG Technical Report (SIGBIO38).

Technology keywords:

  • C/C++
  • Python
  • Java
  • MPI/OpenMP
  • Job Scheduling
  • Pipeline Development
  • The K computer
  • Bioinformatics
  • HPC Engineering

Bachelor of Engineering

Department of Computer Science, Faculty of Electro-Communications,
The University of Electro-Communications
2011 - 2013
Dept. of Computer Engineering, National Institute of Technology, Toyama College
2006 - 2011

Publications

Peer-reviewed Journal Paper

  • Kento Aoyama, Masanori Kakuta, Yuri Matsuzaki, Takashi Ishida, Masahito Ohue, Yutaka Akiyama, "Development of Computational Pipeline Software for Genome/Exome Analysis on the K computer", Supercomputing frontiers and innovations, 7(1), 37-54, 2020. [doi: 10.14529/jsfi200102]
  • Kento Aoyama, Masanori Kakuta, Yuri Matsuzaki, Takashi Ishida, Masahito Ohue, Yutaka Akiyama, "Development of Exome Analysis Pipeline on the K Computer", IPSJ Transactions on Advanced Computing System(ACS), 9(2), 15-33, 2016. [PDF]

Peer-reviewed Conference Paper

  • Kento Aoyama, Hiroki Watanabe, Masahito Ohue, Yutaka Akiyama, "Multiple HPC Environments-Aware Container Image Configuration Workflow for Large-Scale All-to-All Protein-Protein Docking Calculations", Supercomputing Asia 2020 (SCA'20), Singapore, Feb. 26, 2020. [doi: 10.1007/978-3-030-48842-0_2]
  • Kento Aoyama, Yuki Yamamoto, Ohue Masahito, Yutaka Akiyama, "Performance evaluation of MEGADOCK protein-protein interaction prediction system implemented with distributed containers on a cloud computing environment", The 25th International Conference on Parallel and Distributed Processing Techniques and Application (PDPTA'19), Las Vegas, Nevada, US, Jul. 29, 2019.

Conference Paper (without peer-review)

  • Kento Aoyama, Masahito Ohue, Yutaka Akiyama, "Survey and Performance Comparison of Container Technologies in Bioinformatics Field, IPSJ SIG Technical Report, 2017-BIO-54(47), pp.1-7. 2018.
  • Kento Aoyama, Yuki Yamamoto, Masahito Ohue, Yutaka Akiyama, "Evaluation of Container Virtualized MEGADOCK System in Distributed Computing Environment", IPSJ SIG Technical Report, 2017-BIO-49(3), pp.1-8. 2017. [Slide]
  • Kento Aoyama, Masanori Kakuta, Yuri Matsuzaki, Takashi Ishida, Yutaka Akiyama, "Large-scale Parallelization of Exome Analysis Pipeline on K-computer", IPSJ SIG Technical Report, 2014-BIO-38(33), pp.1-7. 2014.
  • Kento Aoyama, Kyouhei Akai, Junichi Akita, Akira Tsukada, "Improvements of the GPU accelerated Simulation for Mach Bands Generation Mechanism in Human Visual System Model", IEEJ Hokuriku Branch, B-5, 2010.

Poster Presentation

  • Kento Aoyama, Hiroki Watanabe, Masahito Ohue, Yutaka Akiyama, "Multiple HPC Environments-Aware Container Image Configuration for Bioinformatics Application", Technical Program (Research Poster) in SC19: The International Conference for High Performance Computing, Networking, Storage, and Analysis, Denver, Colorado, US, Nov. 2019. [Session]
  • Kento Aoyama, Masahito Ohue, Yutaka Akiyama, "Comparison of the HPC Container Technologies for Bioinformatics Application", Real World Big-Data Computation Open Innovation Laboratory (RWBC-OIL) 1st Workshop, May. 2018.
  • Kento Aoyama, Yuki Yamamoto, Masahito Ohue, Yutaka Akiyama, "Large-Scale Parallelization of Cloud Native PPI-System using Container Virtualization", ACLS International Summer School 2017, Jurong West, Singapore, Sep. 2017.
  • Kento Aoyama, Yuki Yamamoto, Masahito Ohue, Yutaka Akiyama, "High-Performance Protein-Protein Interaction Predictions on Distributed Containerized Environment", High Performance Computing Symposium 2017 (HPCS 2017), Jun. 2017.
  • Kento Aoyama, Masanori Kakuta, Yuri Matsuzaki, Takashi Ishida, Yutaka Akiyama, "Large-scale Parallelization of Exome Analysis Pipeline on K-computer", ACLS International Summer School 2014, West Lafayette, Indiana, US, Aug. 2014.

Oral Presentation

  • Kento Aoyama, "Introduction of ML Platform, a Machine Learning Platform in CyberAgent, and recent DevOps issues", Private Cloud Meetup #4, Aug. 2023.
  • Kento Aoyama, "Distributed Computing Service and Platform for LLM Development in CyberAgent", Cloud Operator Days Tokyo 2024 (CODT'24), Closing Keynote, Sep. 2024.
  • Kento Aoyama, Masahito Ohue, Yutaka Akiyama, "Container virtualizations for computational science on HPC environments, Real World Big-Data Computation Open Innovation Laboratory (RWBC-OIL) Annual Meeting 2019, Mar. 2019.

Graduation Thesis

  • Large-scale Parallelization of Exome Analysis Pipeline on K-computer
    Department of Computer Science, Graduate School of Information Science and Engineering, Tokyo Institute of Technology.
  • Large-scale Parallelization of Matrix-matrix Product in Double-double Precision using Fox Algorithm on GPU Cluster
    Department of Computer Science, Faculty of Electro-Communications, The University of Electro-Communications.
  • Improvements of the GPU accelerated Simulation for Mach Bands Generation Mechanism in Human Visual System Model
    National Institute of Technology, Toyama College.