jonkrohn/ml-foundations

This repository holds the code and notebooks for Jon Krohn's Machine Learning Foundations curriculum, a structured series covering the math and computer science topics that sit underneath modern machine learning. The goal is to give learners a grounded understanding of why machine learning algorithms work, not just how to run them through a library. The curriculum is split into eight subjects across four areas. The first two cover linear algebra, including matrix operations. The next two cover calculus, including partial derivatives. Subjects five and six address probability and statistics. The final two cover algorithms, data structures, and optimization. Each subject builds on the previous ones, though the four main areas are independent enough that a learner with existing knowledge in one area can skip ahead. All content is delivered through Jupyter notebooks, which are interactive documents that mix runnable code with text and output. The notebooks are designed to run in Google Colab, a free browser-based environment, so no local setup is required. The material is also available as video courses on YouTube (free), O'Reilly, and Udemy, or as recordings from the Open Data Science Conference. A book based on the curriculum is planned but not yet published. The README explains the motivation with a metaphor: using machine learning tools through their high-level interfaces is the upper floor, but the foundational subjects are what the house is built on. Without them, moving into specialized areas like deep learning or computer vision, where details often exist only in academic papers, becomes harder. The curriculum was originally taught as live online sessions in 2020 and 2021 and is now freely accessible via YouTube. Paid options on Udemy and O'Reilly include exercise solutions and certificates that the free version does not provide.