I built my first personal website in May 2021. I had just entered Kyoto University, studying civil engineering. At the time I was running a geography reference blog called Supergeografi, and I figured hey, maybe I should have my own site too. So I bought a domain from (now defunct) Google Domains. Threw together a Gatsby starter, deployed it to Netlify. It had an about page describing a student who hadn't picked a specialization yet, two project cards, and exactly one blog post: "Hello world! welcome to my personal website."
That was the last thing I ever wrote on it.
The site is still live today, frozen in 2021. It just sat there for five years, through my entire undergrad, through my move to Tokyo for grad school. I never went back to fill it in.
It's not that I stopped caring. I had a whole Notion workspace full of design references, layout sketches, color palettes, typography pairings. I knew exactly what I wanted the site to be: somewhere to share my thoughts, put up work I care about, and present it all in a design language that actually felt like mine. The vision was sharp. I just couldn't get from "here's what I want" to "here's the code that does it."
I'm a transport planning student. I write Python and run spatial analysis. Not exactly a frontend engineer. Every time I tried to turn a layout idea into CSS, the result looked like a homework assignment. That gap between what I imagined and what I could actually ship was pretty demoralizing. And perfectionism made it worse. Every new attempt started from scratch because the last one wasn't good enough. New framework, new design system, new folder structure. The site became a permanent someday project.
What changed
Sometime in late 2025, something shifted in how AI tools worked with code. Not the chatbot-answers-a-question kind, but the kind where you describe a component and it writes the implementation. Where you paste a screenshot, say "make it look like this," and it mostly does.
People called it vibe coding. The term is a bit silly, but it captures something real: you express intent and the machine handles the translation. The gap between "I want a condensed sans-serif heading with tight letter-spacing and a subtle dot grid background" and actual working TailwindCSS classes collapsed to nearly zero.
For someone like me (strong opinions about design, weak muscle memory for implementing it), this changed everything. I could finally work at the level of decisions instead of syntax.
I started the current version of this site in early 2026. Within days I had a working homepage with the exact editorial feel I'd been describing in Notion pages for years. Within weeks I had route transitions, dark mode, a project showcase layout, and even hooked up the Google Books API for a reading shelf.
This is the first version of my site I'm not embarrassed by. It actually looks like what I had in mind.
What this site is for
I'm a grad student at the University of Tokyo, studying transport planning and urban policy. I'm building a career in transport-focused development consulting: analyzing transit systems, modeling freight corridors, writing policy recommendations for governments and multilateral agencies.
This site is where I show that work. The projects section has case studies with real data, real maps, and real findings. The notes section (where you're reading this) is for essays and whatever else happens to be on my mind.
What's next
This post is the first real entry replacing the placeholder content that shipped with the initial build. Up next: actually showcasing the projects I've done. Findings from my bachelor's thesis, a map viewer tool I built during my part-time work, and more to come.
Five years of thinking. A few weeks of building. The site is finally live, and the backlog is finally a pipeline.