Kazuhiro Serizawa

Japanese Software Engineer

Skills

Architecture
  • Microservices
  • Clean Architecture
  • GraphQL
  • RESTful API
Languages, Operating Systems & Tools
  • Ruby
  • Java
  • Kotlin
  • Python
  • JavaScript
  • C++
  • C
  • Rust
  • Terraform
  • Ubuntu
  • CentOS
  • MacOS
  • git
Machine Learning
  • Scikit-learn
  • Chainer
  • Pytorch
  • Jupyter notebook
  • Computer Vision
Platform Development & Administration
  • CircleCI
  • MySQL
  • PostgreSQL
  • nginx
  • fluentd
Data Management
  • MySQL
  • PostgreSQL
  • BigQuery
  • Redash
Containers & Cloud
  • Docker
  • Kubernetes
  • AWS
  • GCP

Projects

chainer_prefetch_multiprocess_iterator (Python)

  • This is the reference implementation of my study “Accelerating Machine Learning I/O by Overlapping Data Staging and Mini-batch Generations”.
  • This is a Chainer Iterator class that executes prefetching training data from slow storages (such like parallel file systems) into fast storage (such as SSD) and generating mini-batches in the same time.
  • The aim of this study is to conceal the time for staging-in training dataset into node-local storages in computation nodes at HPC clusters (such as ABCI, TSUBAME, Cygnus, and so on).

chainer_minibatch_size_optimizer (Python)

inference_engine (C++)

  • This is an ONNX runtime implementation, such like onnxruntime or menoh.
  • For now, Gemm, Conv, MaxPool, Relu, Softmax, Dropout, Reshape are supported.
  • All backend implementation is my own implementation for CPU (This means that current backend implementation does not use optimized matrix libraries, such like Blas, Intel MKL-DDN, and so on.).

simple_map_reduce - Distributed MapReduce framework (Ruby)

optimization_experiments (C)

  • This is an experimental project to study optimization for GEMM and GEMV.
  • This projects includes dgemm and dgemv implementations which are optimized by loop exchange, loop unloop, blcoknize, padding repeatedly.

convolution_experiments (Python)

  • This is an experimental project to study implementation for Convolution with direct and im2col style repeatedly.

config2args (Rust)

  • This is a CLI tool to convert json config file into GNU CLI option style (such like --key1 value1 --key2 value2)

Open Source Contributions

Publications

International conference - Accelerating Machine Learning I/O by Overlapping Data Staging and Mini-batch Generations

The academic paper that I submitted to "the 6th IEEE/ACM International Conference on Big Data Computing, Applications and Technologies (BDCAT'19)"

December 2019

Domestic conference - 大規模機械学習訓練におけるI/O性能の高速化

The academic paper that I submitted to "情報処理学会 第170回 HPC 研究会報告 (HPC170)"

July 2019

Docmestic conference - 深層ニューラルネットワークにおける訓練高速化のための自動最適化

The academic paper that I submitted to "情報処理学会 第168回 HPC 研究会報告 (HPC168)"

March 2019

Experience

Software Engineer, ML

ELYZA, Inc. (株式会社ELYZA)

Developing LLM inference platform

October 2024 - Present

Software Engineer, Mercari Hallo

Mercari, Inc. (株式会社メルカリ)

Developing the GraphQL Server and gRPC server for the Mercari Hallo service

October 2023 - September 2024

Software Engineer, Search team

Mercari, Inc. (株式会社メルカリ)

Developing gRPC severs for search system

March 2023 - September 2023

Software Engineer, Souzoh

Souzoh, Inc. (株式会社ソウゾウ)

Developing gRPC severs for Mercari shops (EC site)

August 2022 - March 2023

Software Engineer

CADDi, Inc. (キャディ株式会社)

Developing GraphQL servers by Kotlin, gRPC servers by Rust, web UI by Next.js + TypeScript, and so on.

June 2021 - April 2022

Senior Software Engineer

Treasure Data, Inc. (トレジャーデータ株式会社)

Developing of internal RESTful Web API systems (Kotolin + Dropwizard), Worker system (Ruby), Workflow Engine (Java8), and so on.

February 2020 - May 2021

Software Engineer (Intern)

Treasure Data, Inc. (トレジャーデータ株式会社)

Development of new functions for OSS workflowengine Digdag

August 2018 - September 2018

Software Engineer

Toreta, Inc. (株式会社トレタ)

Design, implementation, and operation of RESTful Web APIs for the company’s restaurant reservation service (Ruby on Rails 4.x) and internal ETL system

January 2016 - March 2018

Software Engineer

HeartRails, Inc. (株式会社ハートレイルズ)

Design, implementation, and operation of Web Applications and RESTFull APIs for the customer’s web services with Ruby on Rails (4.x)

July 2014 - December 2015

Software Engineer

BizReach, Inc. (株式会社ビズリーチ)

Design, implementation, and operation of Web Applications for the company’s job hunting service (Java 1.6)

December 2012 - June 2014

Software Engineer

Public Identity Co.,Ltd. (株式会社パブリックアイデンティティ)

Design, implementation, and operation of Web Applications for the cocustomer’s web services (PHP 5.1)

November 2011 - December 2012

System Engineer

NEC Nexsolutions, Ltd. (NECネクサソリューションズ株式会社)

System operation for customers and project management support

April 2009 - December 2011

Education

University of Tsukuba, Depeartment of Computer Science, Graduate School of Systems and Information engineering (筑波大学大学院システム情報工学研究科コンピュータサイエンス専攻博士前期課程)

Master of Engineering
I have researched about methods to accelerate the read I/O part at Deep Learning training on HPC Clusters. My master’s thesis title is “Research for Acceleating Large Sclae Machine Learning” (大規模機械学習の高速化に関する研究).

April 2018 - March 2020

University of Iwate, Faculty of Agriculture (岩手大学農学部農林環境科学科)

B.S. in Agriculture
My major was Environmental control engineering. My bachelor’s thesis title is “Development of a Classification Model using Image Information for Determining the Maturity Level of Gentian Flowers” (切り花リンドウの成熟度判別における画像情報を⽤いた分類モデルの構築).

April 2006 - March 2009
Nifty tech tag lists from Wouter Beeftink