The first seven years of my career were spent mostly fixing crashes in a mainframe OS. When I first came to California, there were two piles of crash dump printouts six feet high waiting for me. Gradually I worked down the pile, and time between crashes went from hours to months.

Then I got into high-security operating system development for DoD, which led to proof of correctness, networking, and other theory. The aerospace company was happy to hire someone who'd been deep inside a major operating system and had fixed that many crash-type bugs. I was happy to pivot to the problems of designing reliable and secure systems.