Recently I've been trying to "systemize" my learning a bit. In general, my high-level plan is to learn the all major branches of science to a roughly undergrad level (go deeper or shallower selectively), focusing on the fundamental, core stuff.

This probably sounds ambitious, but you can use the Pareto principle (get 80% of the value by spending 20% of the time) and do various optimizations such as taking care to choose the best learning resources.

I have a learn.md file in Dropbox. It contains several sections: math, statistics, machine learning, physics, chemistry, biology, economics, programming. Each section has a list of subfields, for example biology has "molecular biology of the cell" (that's a book) and "brain". Each section also has a "resources" list and a "topics" list, which contains little topics I'm curious about, for example the math sections has svd and wavelets. And there's also a "next" section containing the list of things I plan to learn next.

Having an overview/roadmap like this really helps with the anxiety about what to focus on next, it makes me feel organized.