I was a PhD student for a while, but I always enjoyed "refactoring" proofs, more than coming up with new ones. Making them simpler, shorter, clearer. Unfortunately, it's not much rewarded.