Yuki Matoba, Developer in Tokyo, Japan
Yuki is available for hire
Hire Yuki

Yuki Matoba

Verified Expert  in Engineering

Bio

Yuki is a full-stack MLOps and DevOps engineer with over five years of experience working for various companies, from startups to large companies. Yuki's background as a machine learning engineer and a software developer helps him understand and solve real-world machine learning and software development problems.

Portfolio

Freelance
Python, Amazon SageMaker, Google BigQuery, Kubernetes, AWS Step Functions...
Cinnamon
Amazon SageMaker, Amazon EKS, Kubernetes, Python 3, Cisco Meraki...
LINE Corp.
Python 3, Kubernetes, TensorFlow, C++, Docker, Amazon EKS...

Experience

  • Python - 8 years
  • Python 3 - 5 years
  • Scikit-learn - 5 years
  • Amazon SageMaker - 4 years
  • Amazon EKS - 4 years
  • Amazon Elastic Container Service (ECS) - 3 years
  • Kubernetes - 3 years

Availability

Part-time

Preferred Environment

Amazon Web Services (AWS), Kubernetes, TensorFlow, Data Build Tool (dbt), FastAPI

The most amazing...

...thing I've done was apply SageMaker for training and Seldon for inference to the ML team and save 70% on model training costs and fluent deployment flow.

Work Experience

DevOps | MLOps | Machine Learning | Software Engineer

2020 - PRESENT
Freelance
  • Worked for Plotly and was in charge of DevOps/infrastructure of Dash Enterprise 5.0. I added GPU/Rapids AI support to the Kubernetes cluster of Dash Enterprise and worked on CI/CD pipeline with vCluster, ArgoCD, and Github Actions.
  • Contributed to Woven Alpha Inc. (Toyota Research Institute, Advanced Development Inc) and refactored the hyperparameter tuning system made with AWS Batch, weights and biases, and step functions and implemented the labeling system's conversion scripts.
  • Developed a large ETL system to deal with training data on AWS for Woven Alpha, Inc. (Toyota Research Institute, Advanced Development Inc).
  • Built models and infrastructure for MiddleField Inc. to offer personalized items using Amazon Personalize and SageMaker; implemented the model infrastructure environment to predict prices of used cars by using Kubeflow pipelines and Seldon Core.
  • Constructed models to predict who leaves companies for AI CROSS and built the environment to develop and evaluate models using MLFlow, ECS Fargate, and Kedro.
  • Built a KPI tree and improved it by analyzing data with SQL and implementing the new algorithm in API developed by Ruby on Rails for React, Inc.
Technologies: Python, Amazon SageMaker, Google BigQuery, Kubernetes, AWS Step Functions, AWS Batch, Amazon Elastic Container Service (ECS), TensorFlow, Google Cloud Platform (GCP), Ruby, Ruby on Rails 4, Docker, AWS CloudFormation, Amazon Web Services (AWS), CI/CD Pipelines, SQL, DevOps Engineer, AWS DevOps, DevOps, Grafana, MySQL, Prometheus, Machine Learning, ETL, Amazon S3 (AWS S3), Data Science, GitHub, Google Cloud, GBM, GitLab, GitLab CI/CD, Continuous Integration (CI), Ansible, MongoDB, Continuous Delivery (CD), Site Reliability Engineering (SRE), Ruby on Rails (RoR), NGINX, Cloud, React, TypeScript, Amazon EC2, Jenkins, Redis, Node.js, Continuous Development (CD), Build Pipelines, GitHub Actions, API Design, JavaScript, Amazon Virtual Private Cloud (VPC), Docker Compose, Helm, Data Build Tool (dbt), Dagster, FastAPI

Infrastructure Manager

2019 - 2020
Cinnamon
  • Built a SaaS product with ML (machine learning) models on EKS cluster using EFS, CloudWatch, and so on.
  • Applied SageMaker for an ML training platform. Posted my work on the AWS blog (AWS.amazon.com/blogs/machine-learning/cinnamon-ai-saves-70-on-ml-model-training-costs-with-amazon-sagemaker-managed-spot-training).
  • Designed an ML training platform with EKS, DVC, Seldon, SageMaker, and so on.
  • Managed an intranet network and security in the Tokyo, Vietnam, and Taiwan offices based on ISMS.
Technologies: Amazon SageMaker, Amazon EKS, Kubernetes, Python 3, Cisco Meraki, Information Security Management Systems (ISMS), Docker, AIOps, Amazon Web Services (AWS), SQL, AWS CloudFormation, CI/CD Pipelines, DevOps, DevOps Engineer, AWS DevOps, Grafana, Prometheus, Linux, Optical Character Recognition (OCR), Machine Learning, Amazon S3 (AWS S3), GitHub, Ansible, PostgreSQL, Continuous Integration (CI), Continuous Delivery (CD), Site Reliability Engineering (SRE), Argo CD, SaaS, MongoDB, NGINX, Cloud, Redis, Amazon EC2, Amazon RDS, CentOS, Jenkins, RabbitMQ, Apache Kafka, Windows Server, VPN, Continuous Development (CD), Build Pipelines, GitHub Actions, Docker Compose, Amazon Virtual Private Cloud (VPC), Helm

Software Engineer

2018 - 2019
LINE Corp.
  • Developed Clova, an AI assistant in smart devices—more information can be found at Clova.line.me.
  • Oversaw and was in charge of the NLU and dialog system in Clova—more information can be found at Speakerdeck.com/line_developers/nlu-architecture-and-ml-model-management-in-clova.
  • Constructed an OSS framework to manage ML modules working on Kubernetes—more information can be found at Github.com/rekcurd.
  • Researched and experimented with building a BERT-like lightweight language model.
Technologies: Python 3, Kubernetes, TensorFlow, C++, Docker, Amazon EKS, Amazon Web Services (AWS), SQL, Python, AWS CloudFormation, DevOps, AWS DevOps, DevOps Engineer, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Machine Learning, Data Science, Amazon S3 (AWS S3), GitHub, Terraform, Ansible, Continuous Integration (CI), Microservices, Cloud, React, Travis CI, API Design, JavaScript, PyTorch, Data Build Tool (dbt), FastAPI

Software and Infrastructure Engineer

2015 - 2017
Nyle
  • Developed a search microservice in Scala, Spark, and CloudSearch.
  • Managed AWS as an SRE (site reliability engineer) and architect for all services in the company.
  • Developed a web application for new business in Scala and domain-driven design.
  • Conducted the first stage interviews for new engineers.
Technologies: PHP 7, Scala, Scikit-learn, Amazon CloudSearch, Spark, SQL, Amazon Web Services (AWS), Terraform, CI/CD Pipelines, Web SQL, DevOps, AWS DevOps, DevOps Engineer, Machine Learning, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), GitHub, OpenShift, Continuous Integration (CI), Ansible, PostgreSQL, Site Reliability Engineering (SRE), Microservices, Continuous Delivery (CD), NGINX, Cloud, HAProxy, Redis, Apache, Amazon EC2, Amazon RDS, CentOS, Load Balancers, Postfix, Node.js, Build Pipelines, Continuous Development (CD), API Design, JavaScript, Amazon Virtual Private Cloud (VPC), Docker Compose

SageMaker training environment

I applied SageMaker to training infrastructure of our company.
SageMaker spot training costs much less than our own system, and it is easy to manage server resources and access permissions.

https://aws.amazon.com/blogs/machine-learning/cinnamon-ai-saves-70-on-ml-model-training-costs-with-amazon-sagemaker-managed-spot-training/

Framework to Manage ML Models on Kubernetes

https://github.com/rekcurd/
Rekcurd is a software package for the management of machine learning (ML) modules. Rekcurd makes it "easy to serve ML module," "easy to manage and deploy ML models," and "easy to integrate into the existing service." Rekcurd can be run on Kubernetes.

Presentation | NLU and Dialog System of Smart Speaker

https://speakerdeck.com/line_developers/nlu-architecture-and-ml-model-management-in-clova
I made a presentation about the NLU and dialog system of Smart Speaker. I architected and developed this by combining ML models and a rule-based model.

In this presentation, I explained the whole architecture and how to build, update, and deploy ML models with less effort in terms of MLOps and DevOps
2011 - 2015

Bachelor of Science Degree in Computer Science

Ohio Northern University - Ada, OH, United States

FEBRUARY 2020 - FEBRUARY 2023

AWS Certified Solutions Architect Associate

AWS

Libraries/APIs

Scikit-learn, Node.js, React, TensorFlow, PyTorch

Tools

Amazon SageMaker, Amazon EKS, AWS CloudFormation, Terraform, GitHub, Ansible, Docker Compose, Amazon Virtual Private Cloud (VPC), AWS Step Functions, AWS Batch, Amazon Elastic Container Service (ECS), Grafana, Jenkins, VPN, Helm, Cisco Meraki, GitLab, GitLab CI/CD, NGINX, Apache, Postfix, RabbitMQ, Travis CI

Languages

Python 3, Python, SQL, JavaScript, C++, Ruby, TypeScript, PHP 7, Scala, Lustre

Paradigms

Web Architecture, DevOps, Continuous Integration (CI), Continuous Delivery (CD), Continuous Development (CD), ETL, Microservices

Platforms

Docker, Amazon Web Services (AWS), Linux, Amazon EC2, Kubernetes, Google Cloud Platform (GCP), OpenShift, CentOS, Apache Kafka, Windows Server

Storage

Web SQL, MySQL, Amazon S3 (AWS S3), Redis, Google Cloud, PostgreSQL, MongoDB

Frameworks

Ruby on Rails 4, Ruby on Rails (RoR), Spark

Other

Software Deployment, AIOps, CI/CD Pipelines, DevOps Engineer, AWS DevOps, Machine Learning, Data Science, Site Reliability Engineering (SRE), Cloud, Load Balancers, FastAPI, Information Security Management Systems (ISMS), Google BigQuery, Natural Language Processing (NLP), Prometheus, GBM, HAProxy, Build Pipelines, GitHub Actions, API Design, Generative Pre-trained Transformers (GPT), Data Build Tool (dbt), Dagster, Amazon CloudSearch, Optical Character Recognition (OCR), Argo CD, SaaS, Amazon RDS

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring