AI Take Home Project Guide:
What Evaluators Actually Want

Match the stated time estimate plus 20-30% for documentation. If they say 4 hours, spend 4-5 hours. If no estimate is given, ask. Don't spend 20 hours on a 4-hour project—it signals poor time management. A clean, complete solution within the time budget beats an over-engineered one that took triple the time.

Priority order: (1) Does it work and meet requirements? (2) Is the code clean and readable? (3) Are decisions explained in the README? (4) Does it handle errors gracefully? (5) Are there tests? Most candidates fail on #1—they submit incomplete solutions. Meeting all requirements is more important than impressive architecture.

Common failures: (1) Missing requirements—read carefully, (2) No README or setup instructions, (3) Code that doesn't run—test your submission fresh, (4) Over-engineering beyond the scope, (5) No error handling, (6) Hardcoded API keys in the code, (7) Missing edge case handling. Simple, working, documented code beats complex, broken, undocumented code.

Include: (1) Setup instructions—someone should run it in 5 minutes, (2) Architecture overview—explain your approach, (3) Trade-offs—what you chose and why, (4) What you'd improve—shows self-awareness, (5) Time spent—demonstrates integrity. The README is often the first thing evaluators read. Make it clear and professional.

Stick to requirements—don't add features they didn't ask for. Match complexity to time budget. If you have 4 hours: simple architecture, basic error handling, minimal tests. If you have 8+ hours: more robust design, better test coverage. Ask yourself: 'Does this requirement exist?' If not, skip it. Document what you'd add with more time instead of building it.

This is intentional testing of prioritization. Focus on core requirements first, then enhance if time permits. Document what you'd add with more time. Submitting a working core solution with a clear roadmap beats submitting an incomplete ambitious solution. If truly unreasonable, ask for clarification—good companies appreciate the question.

It probably isn't, and you should hold the money. Most cohort dropouts are people who couldn't articulate what they were shipping when they signed up. That's why the consult exists, and why I turn down most applications. If we get on the call and you can't tell me what you'll have shipped at the end of week 8, I'll point you to the AI Native Engineer community until you can.

Yes, the cohort works for people shipping their first serious AI system whether the goal is to land a senior role or to launch a product. The shipped system serves both equally well.

You need to be able to code in Python or TypeScript. Complete beginners can follow the classroom they get access to before the cohort sessions to come in well-prepared.

It's a four-figure investment that we discuss during the 30-minute consult, alongside whether the cohort is the right fit for your project.

Yes, most attendees do. The live session is one Tuesday a week and the async work fits around your existing schedule, as long as you can carve out roughly 6 hours a week.