In the beginning of your career I think you just have to choose problems and persist on them until you've solved them no matter how long it takes (within reason

In the beginning of your career I think you just have to choose problems and persist on them until you've solved them no matter how long it takes (within reason). You learn from people and web research but also by doing and remembering. Persist on problems several times and you'll have started to build some intuitions on where problems hide. You'll have to persist less and less with time, but persistence almost always pays off (or you find out why a solution is impossible, which can also be interesting and useful). Eventually you become a guru. I think this is the traditional route to learning. By trying. Learning from others is another route, but trying is really the basis of learning. Here's a trick you might find useful when you are really stuck and don't see any path forward: change some random related things in the code. If the code does not respond to those changes as you expected then that is a clue! There is something there you don't understand and the unexpected behavior is a clue to it. It's kind of like writing unit tests, checking for what you expect to be true. Keeping a journal is essential also.