My personal site: a short introduction, a blog about how big systems actually work, and a four-in-a-row opponent that learns from everyone who plays it.
Live at tunar.dev.
| Framework | Next.js 16 (App Router, Turbopack), React 19, TypeScript |
| Styling | Plain CSS Modules over a custom-property token layer |
| Type | Geist Sans, Newsreader |
| Data | Supabase Postgres for the game record, Supabase Realtime for friend matches |
| Model | TensorFlow.js to train, hand-written inference and alpha-beta search in the browser |
| Tooling | Biome, Husky, commitlint, GitHub Actions |
| Hosting | Vercel, with Vercel Analytics |
bun install
bun devFor the Supabase-backed parts, copy .env.example to .env and fill it in, then apply supabase/schema.sql
to your project.
bun run build # production build
bun run typecheck # tsc --noEmit
bun run lint # biome
bun run test # bun test
bun run analyze # bundle report in .next/analyze
bun run train # retrain the model into public/model/weights.jsonVercel, from the default branch. Set NEXT_PUBLIC_SUPABASE_URL, NEXT_PUBLIC_SUPABASE_ANON_KEY, and
SUPABASE_SERVICE_ROLE_KEY in the project, point the domain at it, and that is the whole deployment.
content/blog/ posts, one Markdown file per slug
public/model/ trained weights, generated by bun run train
scripts/train.ts the training run
src/app/ routes, metadata, sitemap, feed, OG images
src/components/ UI, the game, the SVG figures, the scene generator
src/lib/ posts, site constants, sounds, Supabase clients
supabase/schema.sql tables and indexes
Read CONTRIBUTING.md before opening a pull request.
MIT.