>You start by rejecting those PRs, saying "write more maintainable code, not quick hacks".

How do you go about that when for example, my previous employer just allowed any software developer to commit to any branch, and there was never any code review happening?