My name is Arata Furukawa, written as "古川 新" in Japanese characters. I am Japanese and have been residing in Tokyo since birth. Currently, I work as a Cloud Native MLOps Engineer at LY Corporation (LINEヤフー株式会社).
I have specialized skills in both data scientific and software engineering fields. I am adept at planning and executing performance-driven and innovative projects from a cross-functional and holistic standpoint. My expertise spans various areas, including building machine learning models from scratch using low-level libraries such as TensorFlow, optimizing large-scale distributed data processing in data center class environments, and developing enterprise-grade scalable MLOps systems. Additionally, I have experience in conducting the operation of large-scale Kubernetes systems, or assuming leadership roles such as product owner and project lead.
From 2017 to 2022, I held the position of Lead AI Engineer at DMM. DMM, a unicorn company renowned for its the dozens of consumer services in Japan. Joining as a Lead Engineer at the inception of the AI Business Unit, my responsibilities encompassed the entire ML project lifecycle—from initial business issue analysis and planning to machine learning model development, system implementation, and ongoing operations.
In 2019, prior to the widespread adoption of the MLOps concept, I planned and developed an MLOps platform utilizing Kubernetes. Developed earlier than open-source software like Kubeflow, this platform, featured advanced automation to address the scalability needs of the organization and facilitate the implementation of technically challenging projects. Through automation, scientists were able to conduct more experiments, achieving scalability to accommodate the organization's expansion. Moreover, scientists to maximize performance and further enabled the development of advanced models. Consequently, it became a standard in running machine learning initiatives within the company.
Having transitioned to Yahoo! Japan in 2022, I currently hold the position of Lead MLOps Engineer at LY, following the merger of Yahoo! Japan and LINE in 2023. In addition to my engineering role, I undertake various responsibilities, including those of a product owner, manager, and project lead, across multiple initiatives. Within the AI Platform team, my primary focus revolves around the development and operation of the Kubernetes environment, a ML platform supporting numerous services within our company. Beyond technical endeavors, I actively contribute to optimizing MLOps processes and lead initiatives aimed at advancing MLOps within the organization. I am also actively involved in operating and fostering the growth of an internal community dedicated to this purpose, which consists of several hundred members.
Moreover, through my concurrent involvement in the commerce division, I fulfill a dual role that transcends organizational boundaries, allowing me to bridge the platform and business domains. This unique positioning provides me with insights from various perspectives, including those of the business side, non-scientist roles, and beyond, enabling me to accurately grasp the multifaceted aspects of MLOps initiatives and approach them from a holistic company-wide perspective. Leveraging my distinctive skills and extensive experience, I lead the conceptualization, proposal, and execution of solutions aimed at enhancing organizational outcomes through MLOps practices.
For instance, while actively engaged in ML projects within the commerce division, I identified challenges in data quality management. I addressed by developping ACP Data Quality, an advanced system that enhances the reliability of model operations, on AI Platform. This solution not only resolved immediate issues but also played a pivotal role in our company-wide MLOps strategy, facilitating scalability and efficiency across the organization. You can view the presentations about this initiative here:
- データ品質をコード化！AIプラットフォームのデータ品質管理システム - LINEヤフー Tech Blog (Dec 25, 2023)
- データ品質をコード化！ LINEヤフーのMLOpsを最適化する "ACP Data Quality" の紹介 - Speakerdeck (Mar 28, 2024)
- 第7回 Data-Centric AI勉強会 - YouTube (Mar 28, 2024)
I also contribute to a team dedicated to promoting the adoption of Go and Python languages within the company. Comprising members particularly proficient in these languages, our team actively shares standard practices and the latest information, as each language specialist.
My aspiration is to create a new era where everyone lives with AI. During my high school years, I ported TensorFlow and developed an on-device inference application for Android smartphones. It has long been my dream to usher in an era where everyone's smartphone becomes AI-enabled.
Surpassing my expectations, the advancement of LLM brings that era within reach.
While there have been instances in recent years where AI has led to business success, it remains a world with more failures than triumphs. However, this is not due to limitations of AI but rather the inadequate readiness of those utilizing it. I firmly believe that those who challenge AI the most, fail with AI the most, and accumulate the most experience with AI will be the ones to realize innovation through AI.
My enterprise experience is a stepping stone to my next endeavor. Through numerous experiences of both failure and success, I aim to transform businesses and human activities, shaping a new era. I seek employment where I can fearlessly embrace failure and continue to embrace challenges.
Thank you for taking the time to read until the end.