Pick a small project that you care about.
It has to be something you care about because you will pay attention to whether the results do something meaningful, which is the key to guiding you toward actually learning what you need to learn. For me, this meant making video games. For my friend a physics educator, it might mean generating random quiz questions with meaningful or realistic numbers.
It also has to be simple. For me, I started a long time ago with a text based game where you fight a sequence of monsters by basically rolling dice with tweaks and theme that mattered to me. For my friend, it might be literally randomly generating 3 random numbers and putting them into a pre-written physics problem.
There is always a way to add on to a project you care about, so start super simple and make it work really well (just meaning, to your satisfaction). Then add on another thing and try to integrate it, and so on.
My CS students when I was TAing always learned the most from long term projects that had enough freedom for them to make it fun for themselves. This is in contrast to curated sequences of academic exercises and expository text.
Some people love this phase of learning the best because it can be the most exciting part ("I hadn't even heard of topic X this morning but now I know enough to be dangerous"), and some just accept this as the slog that's required to get you up to speed enough to build useful things which is the part they enjoy the most. (I'm the latter.) But either way, that's how all of us start.