Fullstack engineer with 9+ years building web and mobile products, most of it as the first or founding engineer at early-stage startups in San Francisco and Tokyo. My depth is on the frontend — React, Next.js, React Native, TypeScript, GraphQL — but I've owned the full stack throughout my career: Go and Node.js services, Firebase, and API integrations with third-party carriers and platforms. I'm comfortable being the person who takes a product from an empty repo to something in users' hands, and equally comfortable joining an existing codebase and improving what's there. At Pathpoint I cut redundant HTTP requests by ~40% by digging into metrics and enabling Apollo Client batching; at EPISOD I reduced API calls by 30% and fixed a failing check-in flow with a retry mechanism. For the past two years I've been building ChalkLab, an AI-powered lesson-planning tool for English teachers. That work has made me practical about LLMs in production rather than theoretical: multi-model routing, prompt caching and batch APIs for cost control, structured output design, and instrumenting products so you can actually tell whether they work. I'm bilingual in Japanese and English and have worked across US, European, and Japanese teams. Currently open to freelance and contract engineering work.…Fullstack engineer with 9+ years building web and mobile products, most of it as the first or founding engineer at early-stage startups in San Francisco and Tokyo. My depth is on the frontend — React, Next.js, React Native, TypeScript, GraphQL — but I've owned the full stack throughout my career: Go and Node.js services, Firebase, and API integrations with third-party carriers and platforms. I'm comfortable being the person who takes a product from an empty repo to something in users' hands, and equally comfortable joining an existing codebase and improving what's there. At Pathpoint I cut redundant HTTP requests by ~40% by digging into metrics and enabling Apollo Client batching; at EPISOD I reduced API calls by 30% and fixed a failing check-in flow with a retry mechanism. For the past two years I've been building ChalkLab, an AI-powered lesson-planning tool for English teachers. That work has made me practical about LLMs in production rather than theoretical: multi-model routing, prompt caching and batch APIs for cost control, structured output design, and instrumenting products so you can actually tell whether they work. I'm bilingual in Japanese and English and have worked across US, European, and Japanese teams. Currently open to freelance and contract engineering work.WWWW…