About

I’m Naoki Sega, a software engineer with 20+ years building and scaling the systems behind high-traffic products, and today I’m focused on the cloud and platform infrastructure that powers AI and LLM applications: the cloud-native foundations that make AI and large models practical to run reliably, observably, and cost-effectively. My foundation is cloud-native architecture: 7+ years designing and operating microservices on Google Cloud, Kubernetes, and Go, now paired with hands-on deep-learning study so I can bridge proven engineering practice with the demands of modern AI systems.

The through-line is simpler than it sounds. The scheduling, networking, observability, and reliability work that keeps a large system standing is exactly what production AI systems are built on, so the experience maps almost directly onto what modern AI infrastructure demands.

That foundation runs deep. Before going all-in on cloud-native, I spent 12+ years across the full stack (Java and PHP, with SQL and NoSQL stores: PostgreSQL, MySQL, MongoDB, Redis, and Memcached), and I’ve built and operated backend services for products serving 20M+ monthly active users, with reliability treated as a feature rather than an afterthought. A lot of that work was unglamorous in the best way: tuning application logic, SQL queries, and server configurations to clear out bottlenecks before they reached users.

I’ve led, too. I’ve been the Tech Lead of a 5-person team driving feature development, and earlier in my career I managed engineering organizations of 20+ people. These days I also contribute to the Kubernetes project, where I focus on test and reliability improvements to the scheduler, and I’ve been deepening my ML foundation through a Deep Learning Specialization to keep that engineering-to-AI bridge solid.

A few things I bring

• Cloud-native depth: 7+ years of microservices on GCP, Kubernetes, and Go, with reliability treated as a feature
• High-traffic scale: backend services for products at 20M+ monthly active users
• Full-stack foundation: 12+ years across Java/PHP and SQL/NoSQL, plus performance tuning that makes systems fast and predictable
• Open source: contributions to Kubernetes, focused on test and reliability improvements to the scheduler
• Mission-driven collaboration: shipping high-quality features fast by working closely with PMs, mobile, front-end, ML engineers, and QA

All of it points in one direction. I build things because I want them to make people’s lives genuinely better, and that’s the part of this work I care about most. It’s why I love shipping alongside the people who turn an idea into something real: product managers, iOS and Android engineers, front-end and ML engineers, and the QA folks who keep us honest. The best features I’ve worked on came out of that kind of close, cross-functional collaboration, where everyone is pulling toward a product people actually want to use.

About this blog

Posts here tend to be long-form and hands-on, the kind of write-up I wish I’d found before hitting a problem myself. If something I wrote saved you time, or you spotted a mistake, I’d genuinely like to hear about it.

Full work history and current certifications are on LinkedIn. You’ll find that and my other links in the header.