How the Work Works

This lesson is about how you actually learn it, because the structure of the work is deliberate. One of the biggest lessons I have learned as an engineer is that real learning happens when you hit a problem and work your way through it.

The lessons introduce the ideas and show the techniques. The quizzes, projects, and capstone are where you put them into practice — the way you would on the job.

Three kinds of work

Throughout the courses, you will complete three types of work:

• Quizzes check your understanding of a few lessons while the concepts are still fresh.
• Projects give you guided practice applying a module’s concepts to a complete deliverable.
• The capstone is the one nobody scopes for you: you pick the problem, decide whether it even needs a model, and take it the whole distance.

The amount of guidance changes as you progress. Early work gives you more support; later work gives you more ownership.

Projects

Each module includes one project where you build the type of system that module introduces. The lessons use worked examples to demonstrate each technique; the projects are where you apply those techniques yourself, working with your assigned project dataset.

You carry this project through every module, expanding it as you develop new skills. By the end, you will have a complete system that grew alongside your understanding. Each finished project also goes on your portfolio site — the page you set up in a later lesson of this module — so the work is not just graded and forgotten; it accumulates into a public record you can point an employer at. The upcoming modules return to this, adding each module’s deliverable to the portfolio as you complete it.

The structure reflects how real engineering work happens:

• You build in your own GitHub repository. A finished project that demonstrates your ability to build, explain, and maintain a working system.
• You start from a reliable foundation at each phase. Starter repositories give you a known-good starting point, so you focus on new skills rather than repeating setup.
• You work with real tools and practices. Later projects introduce Docker, APIs, cloud deployment, infrastructure as code, monitoring, and CI, completed in your own environment and cloud account.
• You build each module’s work on its own branch. Rather than changing the main version directly, you create a branch for the module, build there, and merge it back once it is ready — the same workflow professional teams use to keep a working version safe. The first module teaches this in full.
• You submit through your repository. Once your work meets the requirements, you push your changes and submit the repository URL.

Each project page includes the required tasks and a rubric, so you have a clear standard: what should work, what should be included, and which engineering practices should be demonstrated.

The capstone

At the end of the course, you take a problem nobody scoped for you and carry it the whole distance — the same journey the projects walked, on a product of your own choosing.

That phrase, nobody scoped for you, is the whole difference. The projects arrive with a stakeholder’s words already written down and acceptance specs already listed. The capstone starts one step earlier, at the question the job actually opens with: is there a problem here, and is it a machine learning problem at all? You answer that in writing, and then you build what you argued for.

Two things make it harder than the projects on purpose. You choose the dataset, and the suggested ones are far larger than anything the lessons hand you — large enough that how you read the data becomes a decision you have to defend. And you decide which pieces of the journey your problem genuinely needs; skipping one is fine if you say so and say why, and only silence counts against you.

It is graded twice. A free automatic check reads what a machine can honestly read — the documents exist, the tests run, one command trains, the service answers. A paid review is me reading the argument itself, which is the part no checker does and the part interviews probe.

The goal is not to finish one more project. It is to show you can start one — and the capstone becomes the centerpiece of your portfolio site, the end-to-end system that shows in a single place what you can do without instructions.