Start course

The bootcamp capstone

An AI assistant for a real user with a real problem.

Each student builds their own app, so everyone leaves with a portfolio piece they can demo in an interview. They choose the user and the problem on day 1, and every day's lab adds to it.

What the finished app includes9 requirements
  • Live on a public URL
  • Streaming chat or task interface
  • Structured output validated with a schema
  • At least one tool connected to real data
  • Retrieval over real documents, with citations
  • An eval suite with a reported pass rate
  • One agent step or MCP connection to an outside system
  • Guardrails: input checks, PII handling, rate limit, spend cap and secrets
  • A README a stranger can follow
How it's presented and gradedWeek 3 · Tue

Presented on Tuesday of week 3: a 5-minute live demo and 3 minutes of feedback per student from the head teacher and TAs. Fifteen students fit inside the 3-hour session.

It works

25%

Strong · 3

Live at a public link. Handles odd questions from the room without breaking.

Solid · 2

Live and the main flow works. One or two rough edges show up in the demo.

Not yet · 1

Only runs on a laptop, or the main flow breaks during the demo.

Engineering

25%

Strong · 3

Code another student can follow. Evals run and show a pass rate. Bad input is caught.

Solid · 2

Code is readable. A few evals exist. Some safety checks are missing.

Not yet · 1

No evals, or code only the author can follow.

Product judgment

25%

Strong · 3

A named user with a real problem. The app does one thing for them well.

Solid · 2

A real problem, but the scope is a little wide or the user is vague.

Not yet · 1

No clear user, or the app tries to do too many things.

Communication

25%

Strong · 3

The demo tells a story in time. The README gets someone running. Answers are direct.

Solid · 2

The demo is clear but runs long, or the README misses a step.

Not yet · 1

The demo is hard to follow, or there is no README.

Example ideas4 ideas
  • A course-syllabus assistant
  • An internship application tracker
  • An allergen and menu bot for a local restaurant
  • A knowledge base for a campus club