For getting a job at Google/Facebook/Amazon/X cool startup, it's a lot more important to do some narrow-minded rote-memorization of algorithms and data structures and hone the skills talked about in a small set of books like Cracking the Coding Interview than to learn something really interesting, or become good at a certain thing, or build some cool shit.

Most interviewers aren't really interested in figuring out "is it worthwhile that you built this thing / did you do it right / why do you think this was something worth learning?" when it's so much easier to evaluate your whiteboard code on a narrow problem against some ideal standard solution.

You can tell them about how you were able to learn enough about a particular database in a couple weeks to be able to cut 90% of the runtime out of a hot query in your first few months at a job, since that was an area that wasn't scaling well and it was something your boss asked if you could take a look at, and they'll be like "ok that's nice, now implement a trie." What's the point of even having experience building and running real systems if that's the process?

If you'd rather be building than practicing whiteboard questions, your best path to success is probably going to be a less predictable one where you need to come across one of the right companies willing to work harder at interviewing because they have to find the best candidates who fall through the cracks at the places with the name recognition.