The first time I interviewed at Google I didn’t make it past the phone screen. I bombed a question about an appropriate length of a Bloom filter because I hadn’t recently been thinking about Bloom filters, and I was summarily rejected for onsites. The second time I interviewed I was asked a number of questions that I was perfectly “primed” to answer. Because those were in my brain’s cache, I was able to get through the full interview loop with a “hire” recommendation.

My focus is security, and so the second time around I got Travis Ormandy as an interviewer. I was able to impress him well enough with my IPSec foo to make it onsite. Then I got asked a question I had already been asked in another interview with another company, and I was able to knock that out of the ballpark (I mentioned I already knew that question, but the interviewer wanted me to answer it anyway). Then I got asked a question that used recursion, and because I took a graduate course on induction and recursion within the past year, that was effortless for me. Then I got asked a question that involved a kernel feature, and because I happened to recall the Linux kernel list macros and semantics (something I couldn’t repeat now over 10 years later), I really impressed that interviewer.

I ended up having a really successful part of my career at Google, even though I ended up not really writing much code when all was said and done. However what code I did write was highly impactful.

Looking back, things could have gone very differently if I had gotten a different set of interviewers and/or questions. The questions I got just happened to coincide with what was “paged in” to my “working memory,” so to speak.