NextInML

NextInML is a directory for early-stage machine learning students. It lists summer schools, fellowships, and internships, and gives each one a detail page. Program data comes from Airtable when that is configured, and from a built-in sample set otherwise.

This file covers local development. For how the parts fit together, read docs/architecture.md.

Stack

• TanStack Start (React 19), server-rendered through Nitro.
• Vite 7 for the dev server and the build.
• Tailwind CSS v4 for styling.
• TanStack Router for file-based routes, TanStack Query for client-side data fetching.
• Default build target is Cloudflare Workers.

Prerequisites

A recent Node LTS, or Bun. Bun is set up here (there is a bun.lock and a bunfig.toml). npm works too, since a package-lock.json is checked in. Pick one and stay with it.

Install

bun install   # or: npm install

Run the dev server

bun run dev   # or: npm run dev

![NOTE] The server starts on port 8080. If that port is busy it takes the next free one, so check the startup log for the real URL.

Build and preview

bun run build
bun run preview

Lint and format

bun run lint
bun run format

Environment variables

Program data loads from Airtable when these are set:

• AIRTABLE_TOKEN: a read-only Airtable personal access token.
• AIRTABLE_BASE_ID: the base that holds the Programs table.
• AIRTABLE_TABLE_NAME: optional, defaults to Programs.

Set them in your shell before starting the dev server, or in a gitignored env file your tooling loads (.dev.vars and *.local are already ignored). Without them, the app falls back to the sample programs in src/lib/programs.ts, and the directory shows a note that it is running on sample data. The token is read inside a server function, so it never reaches the browser.

Where things live

• src/routes: pages and the root shell.
• src/components: shared components, with UI primitives under src/components/ui.
• src/lib: the Program type, the Airtable server function, and small date helpers.

The full map is in docs/architecture.md.