A data scientist's day-to-day work usually splits between cleaning messy real-world data, exploring it for patterns, building models to answer a specific business question, and communicating what was found to people who don't have a technical background. The glamorous "training a model" part is often the smallest piece of the job.
The core skill stack is fairly consistent across the field: enough programming to manipulate data and build models, usually in Python or R; enough statistics to know when a result is meaningful versus noise; and enough domain knowledge to ask the right questions in the first place, since a technically perfect model built on the wrong question is worthless.
For beginners, the most efficient path is usually a small number of real, messy projects — using public datasets on a topic you actually care about — rather than working through course after course without ever building something end to end. Employers consistently value a portfolio that shows finished, explained work over a long list of completed tutorials.