Daigo Hiroka

Data & Machine Learning

Career Summary

After working as a machine learning engineer at a data analytics consulting company, I joined CADDi in December 2022 as part of the AI Lab MLOps Team, where I was involved in the development and operation of a machine-learning-based drawing analysis system. From 2023, I also worked as an SRE for CADDi Drawer, handling infrastructure maintenance and the promotion of SLI/SLO operations. In 2024, I moved to the Analysis Platform Group, where I once again focuses on MLOps as well as building inference infrastructure and ensuring reliability for machine learning systems.

Why I Joined

As I considered changing jobs, I noticed that CADDi had started investing seriously in machine-learning-based drawing analysis and seemed to be attracting strong talent as well. One of my key criteria was that I wanted to be involved in building and operating a service that would continue growing globally. I could see that engineers and business-side members at CADDi were working together as one to drive the growth of the business. I felt strong potential in what the company could become and decided that I wanted to dive into that growth myself. Since joining as an engineer on CADDi Drawer and working on areas such as SRE and machine learning infrastructure, I have once again come to appreciate both the difficulty and the excitement of building a scalable service for a growing user base and expanding amounts of data. It is not an easy challenge, but it feels absolutely worth betting my engineering career on.

Role at CADDi

At CADDi Drawer, I contribute to engineering mainly from two angles: SRE and MLOps. As an SRE, I have worked to improve the service's reliability through infrastructure maintenance, the promotion of SLI/SLO operations, incident response, and emergency preparedness such as on-call systems. As an MLOps engineer, I work on the machine learning backend for drawing analysis and use my SRE experience to improve SLI/SLO and other non-functional requirements, all with the goal of building a scalable and highly reliable analysis platform. Looking ahead, we expect further growth in data volume, data types beyond drawings, and the range of analyses we want to support, so I intend to continue building out a scalable machine learning platform that can keep up with that expansion.