Like Kevin I started out tinkering/taking apart everything as a child and reading electronic theory books from the local library. Then we got a computer some time in the mid-90s and I spent hours reading up on various electronics projects/forums online. Up through highschool there were barely any electronics related classes (well there was a tech class but that didn’t really go beyond wiring light bulbs with those spring connection breadboards).
I then went on to attending a top engineering college which after a few semesters I dropped out of. It was a combination of finances, the death of a close family member, and my frustration with academia as a whole which resulted in me taking some time off from education. I then transferred my credits to another engineering college and finished my BSEE. Then I went to yet another school for my MSEng. In grad school I also performed research in a power electronics lab and was a teaching assistant for intro/2nd year analog/digital electronics classes.
Overall, I’d say that computer architecture classes were the most interesting/fun (basically similar to what you’ve described with CMOS logic, register level design, assembly programming). I absolutely hated having to take required courses which were only vaguely related to electrical engineering. Basically physics 1 (EM was ok but still not very relevant imo), english/writing, probability and any math beyond diff eq (I’m looking at you multivariable lol). I think what really killed college for me the first time around was a lot of my first interactions with fellow students and professors were not good at all (there was a lot of arrogance/hubris at my first school and the student to teacher ratio was massive, like 100:1).
Anyway now I currently am a research and design engineer at a company that produces medical ultrasonic equipment. Funnily enough I’ve found that most of what professors hammered into me during school I rarely use and most of what they glossed over is critical. Like rote memorization of specific equations is far less important than being able to do so called back of the envelope ballpark/feasibility calculations.