I grew up in Hà Nội, moved to Japan young, and picked up engineering at Le Wagon. Then I spent four years at KOMOJU, a Japanese payment gateway.

Payments are a good domain for learning careful engineering: mistakes have money attached to them, retries have to be idempotent, and state has to be right when things go wrong, which they will. My work split between backend-heavy full-stack features and internal developer tooling, with constant vigilance for PII protection and auditability. What I like most is the invisible-but-material improvement: the migration that makes a slow tool feel fast without anyone having to relearn it. Living in Japan taught me to see that as craft: the best-designed things don't announce themselves; you use them and don't notice the thought that made them feel natural.

At KOMOJU I was writing less code and reviewing more as AI improved. Now as a student I use it to think alongside me: for sketching, reasoning through tradeoffs, writing code once I know what I'm building. The mental model of the codebase has to be mine, not the AI's. Otherwise the second something breaks in production, I get to enjoy debugging code I don't remember writing.