Exercism is a resource I wish more people knew about, because it is a rare example of something in the coding education space that is both genuinely high quality and completely, sincerely free, run as a nonprofit rather than as a funnel toward some paid upsell. The core of it is straightforward but well executed, which is a large set of coding exercises across more than seventy language tracks that you solve on your own machine in your own editor, running the tests locally and pushing your solution up when it passes, and this local first approach is a quiet strength because it builds the real habits of writing, running, and debugging code in a proper environment rather than the artificial comfort of an in browser sandbox. The feature that genuinely sets Exercism apart, though, is the mentoring. On top of the automated tests, real human volunteers will review your solutions and offer thoughtful feedback on how to make them cleaner, more idiomatic, or more efficient, and getting that kind of individual attention from an experienced coder is the sort of thing people normally pay real money for, so finding it here at no cost is remarkable.
For the audience of this site the relevance is specific and worth being clear about. Exercism does not teach you machine learning or AI, and it does not pretend to, so nobody should come to it expecting neural networks or model training. What it teaches is programming fluency, and the Python track in particular is an outstanding way to get truly comfortable with the language, and since Python fluency is the bedrock that essentially all practical AI and machine learning work is built on, drilling it here until writing code feels natural pays off enormously when you later sit down with a data science course and want to focus on the concepts rather than fighting the syntax. The honest limitations flow directly from what it is.
There is no structured curriculum marching you toward a particular career outcome, so you have to supply your own direction and discipline, the mentoring depends on volunteers and so response times naturally vary, and the exercise based format sharpens your command of language features and problem solving without teaching you how to architect and build a complete application from nothing. None of that is really a criticism, because it is simply not what Exercism is trying to be. My recommendation is warm and simple. If you want to become genuinely fluent in Python, or indeed any language, as a foundation for your AI learning, Exercism is one of the best free tools in existence for exactly that, and the combination of real local coding practice with free human mentoring is something I would happily point any learner toward.
Use it to make the fundamentals second nature, then carry that hard won fluency into a dedicated machine learning path.