I don't know who Brenden Gregg or Jeff Dean are, but a cursory look at their profiles I imagine the thread linking them together is they've set aside time to do deep thinking and research -- rather than the piecemeal approach of implementation whack-a-mole.
So in that vein:
1. Find a very narrow area where you would like to have in-depth knowledge
2. Read the seminal works in that area (could be academic papers, could be blog-posts, could be even code)
3. Go for a long walk, and mull about it for a bit.
4. See if you can add something, no matter how little, to the area
5. ???
6. Rinse repeat
---
Big question is whether that's something you're internally motivated by, or externally. The fruits of that sort of labor are mostly internal; and very few people have the stamina to continually do this, unless it personally fulfills them.