A tiny, shippable growth experiment built on Val Town.

This project is intentionally small: one email, two links, and a fully-instrumented mini web app. It demonstrates how I approach growth engineering at dev-tool companies—ship fast, measure everything, and wire feedback loops end-to-end.

pillChoice is a two-path experience inspired by the classic “choose your pill” metaphor:

• Red pill → a public, role-tailored CV page + concrete growth ideas
• Blue pill → a hidden route that triggers a server-side automation

Both paths log events to a shared SQLite database so behavior is observable immediately.

This was built as an application experiment for a Growth Engineer role at Val Town.

Rather than sending a traditional cover letter, I wanted to:

• Build inside the platform
• Show how I think about activation, instrumentation, and growth loops
• Demonstrate comfort with Val Town’s execution model (Deno, Hono, stdlib)

• Single-page React app

• No router library — routing is done via window.location.pathname

• Client-side event tracking on:

  • page views
  • pill selection

• Hono HTTP app (index.ts)
• Val Town std/sqlite for persistence
• Optional email + webhook triggers on the blue-pill route

Events are stored in SQLite with:

• timestamp
• event type
• pill choice (if applicable)
• path
• user-agent + referrer

You can inspect recent events at:

/api/events?limit=50

These are not required, but enable extra proof points:

• WEBHOOK_URL — POSTs a JSON payload when /blue is visited
• NOTIFY_EMAIL_SUBJECT — subject line for notification email
• NOTIFY_EMAIL_TO — recipient (if supported by plan)

There is no local dev server.

Instead:

vt watch   # sync changes continuously
vt browse  # open the live site
vt tail    # view logs

The deployed site is the development environment.

• Minimal surface area — fewer moving parts, clearer intent
• Observable by default — every meaningful action emits an event
• Fast to iterate — new experiments should take minutes, not days

Thanks for taking the time 🙏

This project is small on purpose. The goal isn’t polish—it’s to show how quickly I can:

• understand a platform
• build within its constraints
• and turn an idea into a measurable experiment

— Andy