Projects¶
Projects are larger than labs and deliberately underspecified — real work does not come with acceptance criteria attached. Each one integrates several volumes and is meant to end up in your portfolio.
Available now
The full capstone — a service from commit to production — is written and available in Volume 13.
What makes a project portfolio-grade¶
A repository with working code is the minimum. What actually impresses a reviewer:
- A README that explains the problem before the solution
- Tests that would catch a real regression
- CI that runs on every pull request and visibly passes
- A design note explaining what you chose and what you rejected
- Honest limitations — what it does not do, and what you would do next
In the field
Interviewers rarely read your code line by line. They read the README, glance at the tests, check whether CI is green, and then ask you to explain one decision. A project with a thoughtful README and three good tests beats a larger one with neither.
Planned projects¶
| Project | Volumes | What it proves |
|---|---|---|
| Service from commit to production | 3, 12 | You can ship and operate software |
| Batch data platform | 4 | You can move data reliably |
| Analytics and experiment report | 5 | You can reach a defensible conclusion |
| ML system with monitoring | 6, 11 | You can operate a model, not just train one |
| RAG application with evaluation | 8, 9 | You can build and measure an AI system |