ai roadmap
For builders who want to understand models, not only call APIs. Python, data wrangling, classical ML, neural nets, evaluation, operations, and a sober look at ethics.
Stages run top to bottom; skills inside a stage can be taken in any order unless a line says otherwise. Locked nodes need their required predecessor first.
Python fluency plus the two maths intuitions everything else leans on.
Query, wrangle and model tabular data before touching a network.
Networks from first principles, then PyTorch you can debug.
Evaluation you trust, operations you can repeat, packaging you can ship.
Where models hurt people, and the checks that stop yours from doing so.