Foundations
Python
The only genuinely non-negotiable prerequisite in the entire Atlas.
Grasp
Every serious AI library, framework and research artefact assumes Python. This is not a preference, it is an ecosystem fact, and arguing with it costs you years.
You need less of the language than you might think. Functions, classes, comprehensions, decorators, context managers, type hints, and genuine fluency with the standard library will carry you through almost all applied AI work. What matters far more than breadth is comfort with the numerical stack that sits on top of it: NumPy arrays and broadcasting, pandas for tabular work, and the mental model of vectorised operations replacing loops.
If you are coming from another language, budget two to three weeks. If you are new to programming, budget three months and do not skip ahead. Every downstream node in this map will be harder in proportion to how shaky this one is.