Phase 0 — Getting Started - The Modelwright Project

Phase 0 is the on-ramp, and it is built to be genuinely welcoming. By the end of it you will have trained, evaluated, and understood real machine learning models, set up a professional workflow, and learned to handle data. This phase is about experiencing machine learning and getting comfortable with it, so that the deeper phases have solid ground to stand on.

That ordering is deliberate. You start with the map of the field, no code at all. Then you set up your tools and train your very first model with a library, a quick win that shows machine learning actually running. Then you learn to work with data, which is where every real project truly begins. Math sits alongside as a reference you reach for when you need it, with a path to build it up if it is rusty. The phase ends with a complete project you carry from a raw dataset to an honest result, entirely with libraries.

The harder, build-it-from-scratch work begins later, once you know what these things are for. You will never be asked to build a mechanism you have not already used.

Modules (work them in order)¶

0.1 — The Landscape of Machine Learning (orientation · concept check) No code. What machine learning is, how AI, ML, deep learning, and LLMs fit together, the types of learning, the end-to-end workflow, and what an ML engineer does.
0.2 — The ML Engineer’s Toolkit & Your First Model (curated · ship gate) Set up a reproducible environment and experiment tracking, learn the scientific Python stack, and train and track your first model with scikit-learn. Produces a reusable project template.
0.3 — Working with Data (hands-on · ship gate) Loading data and formats, exploring and cleaning it with pandas, and splitting it correctly. The everyday skill behind every real project.
0.4 — Math, Just-in-Time (hybrid · concept check) An indexed reference to the math the curriculum uses, plus a foundational track if you need to build it up. A reference you return to, not a wall to climb.

Phase capstone¶

Capstone 0: your first complete end-to-end project. Take a real dataset, explore and clean it, train and compare a couple of scikit-learn models, evaluate them honestly on data they have not seen, and write up what you found. No internals required. This is your first real result from start to finish, and proof that you can already do machine learning before you ever open up how it works inside. Ship it to its own repository, modelwright-phase-0-capstone, the same way you shipped the module projects, so it stands on its own as the first real piece of your portfolio.

Begin with Module 0.1.