Starter skeletons for the capstones in AI Architect Academy — the build-first program that trains experienced software engineers to become AI engineers and architects (Anthropic · AWS · Cloudflare).
These are deliberately empty: the shape the capstone rubric scores, none of the implementation. Fork them, fill them, deploy them, and put the result in your portfolio.
| Template | For | You get |
|---|---|---|
capstone-starter/ |
Engineer capstone (and its Cloudflare deployment target) | src/lib/ portability seams, a threshold-gated eval harness, a cost-model log |
architect-package/ |
Architect capstone | A section-by-section design-package template — constraints, ADR-style rationale, cost at three scales, threat model, compliance mapping, stakeholder one-pager |
- Fork or copy the template you need into a fresh public repo of your own — build there, not inside the course repo.
- Follow the README inside
capstone-starter/(or the prompts under each heading inarchitect-package/PACKAGE.md). - The engineer capstone is done when
node eval/run.mjspasses its threshold gate; the architect package is done when a reviewer can defend every choice from what you wrote.
A senior's capstone should demonstrate judgment under real constraints, not scaffolding effort. The skeleton fixes the shape — portability seams, an eval gate, a cost model, a rationale document — so the work you show is the decisions, not the boilerplate.
The course is itself built the way it teaches: a live Cloudflare Worker + D1 + an AI coach, with spaced-repetition mastery over a backward-designed curriculum (Track 0 mindset → agentic systems → Anthropic/AWS/Cloudflare platforms → integrative capstones → EU/US AI regulation). The build is the curriculum.
→ aiarch.dev · curriculum
Apache-2.0 — use these freely, including for a capstone you show an employer.