Understand basic mathematics necessary for AI(linear algebra, probability, statistics, calculus), then start looking at some basics related to linear regression, logistic regression, naive bayes, decision tree, bagging boosting, SVM & kernels, clustering algorithms, perceptron, back propagation, diminishing gradients problem, weight initialization, RNN, LSTM, attention mechanism, CNN, then from there depending on which path you are interested in, pick models in them and try understanding them.
Also for time series, there are good set of principles that you can look at.

This is only for a scientist career, if not learn the libraries and code what scientists have proposed, try learning to scale, etc. I may not be right person to talk about this side as i have no experience there.