This is a sort of meta-skill I developed during my college years that has served me extremely well up to now. Before I started learning to do this, I would think of project ideas and more or less immediately discard them, since I didn't yet have the relevant skillset to accomplish it and felt that acquiring that skillset would be too much of an investment. Something like "make a machine learning classifier for seeding tournament brackets" felt impossible to me, since I was under the erroneous assumption that I would need to learn mountains of machine learning to "become worthy" of doing a project on machine learning. In short, I felt that I didn't deserve to use tools that I had not yet mastered completely.

Luckily, I realized, that's not really the case! Becoming functional and mediocre at pretty much anything is within reach for pretty much anyone, and if you're doing it for the sake of a project and not a career then it's often practical to only acquire what you need to accomplish your goal. This can often lead you to a more passionate interest in your subject that you otherwise would have had (i.e. making something cool to gauge your interest in a subject, then deciding that it was extremely cool and learning more about it).

The nice thing about this is that your skillset and knowledge don't go away after you finish your project. By doing many things, you learn to do many things, and thereby gain access to even more things you wouldn't have even thought of had you not done the first thing. As they say, once you're given a hammer, nails start popping up all over the place.

It seems like a pretty obvious thing to say, but you won't believe how often I see people place this sort of self-imposed barrier to action upon themselves. People who show up to SSBM tournaments after spending a year practicing by themselves, since they didn't feel "good enough to play with real players", or people who refuse to do anything vaguely math-related since they're afraid of numbers.

Also important to note is that learning in this way will often put you face-to-face with failure. It's important to realize that this is okay! Having consistent output is more important than making sure all your output is good, since you have no obligation to share every single failure you have. Trying something too hard and messing up still nets you experience points in the real world, and as Neal Stephenson put it: "hey, at least a wild-goose chase gives you some exercise."