CS department at my alma mater was very, very theoretical and we were encouraged to just write code in text editors. After that I got into programming for CAD software which had rudimentary editors as well. Basically I just memorized a bunch of stuff and always had a second monitor with the docs open. These days I use autocomplete but I still largely code the same way, and for better or worse I’m set in my ways — at least when I’m using libraries I’m familiar with. If I’m messing around with something completely new then sure, I’ll make extensive use of autocomplete.