keivanlife.com
Half Work, Half Adventure, Total Freedom: a connectivity-aware planner for remote workers exploring Kyushu, Japan.
keivanlife.com- Role
- Founder / Full-Stack Engineer
- Timeline
- 2025–2026 · Primary build Apr–Jun 2025, light maintenance into 2026
- Team
- Solo builder: product, schema, UI, AI planner, maps, auth, and CI/CD
- Stack
- Next.js 15, React 19, TypeScript
KeiVanLife (軽半ライフ) helps digital nomads plan remote-work travel across Kyushu without connectivity surprises. I designed and built the full product end to end: a Next.js App Router web app for workspace discovery, community connectivity reports, and VanAI itinerary planning that schedules work blocks around Wi‑Fi and meetings while leaving room for exploration. The stack is a create-t3-turbo-style monorepo (Neon PostgreSQL with PostGIS, Drizzle, Clerk auth, MapLibre maps, UploadThing photos, and OpenRouter-backed structured itinerary generation) deployed on Netlify.
The problem
Remote workers exploring southern Japan still have to guess whether a café, michi-no-eki, or rural stretch can support a video call. Connectivity data, work-friendly spaces, and trip planning live in separate tools, so people either over-index on tourist hubs or risk dead zones during critical work hours.
KeiVanLife needed a single product surface: contribute and browse connectivity intel, discover remote-work-friendly workspaces, and generate itineraries that treat productivity constraints (meetings, minimum speeds, quiet environments) as first-class inputs, not an afterthought bolted onto a generic travel planner.
My approach
I built the platform as a TypeScript monorepo (pnpm + Turborepo) with Next.js as the BFF: App Router pages, server query modules, and route handlers over a Drizzle schema on Neon with PostGIS enabled for geo-aware data. Clerk handles auth; shared UI lives in an internal design-system package; UploadThing covers workspace and review photos.
The itinerary path is the product differentiator: a multi-step planner captures work schedules, critical meetings, dual timezones, connectivity needs, and travel interests, then VanAI (OpenRouter → Gemini with structured JSON output) returns stops plus backup plans for rain or weak connectivity. Community itineraries, dashboards, profiles, contribution leaderboards, and a markdown blog round out the MVP loop of discover → contribute → plan → save/share.
Product & engineering details
Domain model covers workspaces (cafés, coworking, libraries, hotels, michi-no-eki, and more) with amenities, reviews, and photo galleries; connectivity speed-test reports with voting; itineraries and stops with engagement (likes, bookmarks, saves, views); user preferences for work/travel/connectivity; and persisted AI metadata (model, tokens, cost, latency) for each generated plan.
Maps use MapLibre with Carto basemaps on workspace and itinerary views. Rate limiting runs through Upstash Redis; env validation uses T3 Env; CI/CD is GitHub Actions into Netlify. I kept honesty in scope: the connectivity map UI ships, but some layers still use sample points; subscriptions appear in schema/settings without a live payment provider; Expo mobile and admin tooling remain on the roadmap.
Outcome
keivanlife.com is a substantial solo MVP for connectivity-aware remote-work travel in Kyushu: authenticated users can browse and review workspaces, submit connectivity reports, generate and edit VanAI itineraries, browse community plans, and manage profiles and preferences. The codebase is a coherent full-stack product foundation, ready for live map data wiring, billing, mobile, and ops hardening, and a clear portfolio piece for end-to-end ownership from problem framing through schema, AI, and deploy.