Man, what a great read! As someone currently in the recruiting process with the AI labs, I found myself relating to so much of this. My favorite part was your point about “preparing like an Olympian.” People around me sometimes think it’s insane to dedicate every spare hour (weekends, holidays, and evenings outside your 9–5) to recruiting in the weeks leading up to interviews. But in the end, those few hours can completely change your life trajectory, so why not give it everything you’ve got? Thanks so much for writing this up. It honestly made me feel way better about how I’ve been approaching things.
I suffer from severe confidence issues and was always doubting myself after not getting screened, or flunking an interview. Have been following since your "Work-Life Balance slows careers", and your honest insights have been invaluable. I think it's rare for high achievers to acknowledge how hard they work as well, since it's nice to be called a genius who doesn't ever have to make sacrifices.
Thanks for this. I recently started the intense prep for system design interviews, so this strategy resonates with me. Sadly, it took bombing a few interviews to convince me it’s time to treat interview prep like a full-time job. It’s nice to feel less alone in this!
I'm the opposite. I've given so many system design interviews that they are pretty much second nature at this point. Leetcode type problems still trip me up though.
A bit of a tangent but I had a similar experience with system design interviews (even though I've actually done a bunch of them as an interviewer myself). What does it say about the quality of the interviewing process when one who can admit a glaring weakness in this space and still apparently do a good enough job to be hired with just two weeks of intense preparation (and some luck in problem/interviewer selection) lol. I have some thoughts on this of course but I would be curious what you think on the topic.
Absolutely — your point has profound implications on the whole process of interviewing itself, the premise of it. For instance:
- What does it mean when OpenAI turns down referrals I made where I am _positive_ the referral would have been a stellar addition to the team? I've managed hundreds of engineers and am unlikely to be wrong about this.
- Microsoft insists on interviewing _internal_ candidates when moving between teams. I had many years of 4.5 reviews (a very high rating), but was still declined to join other teams due to 3 hours' interviews.
I agree completely that we aren't using much higher signal mechanisms by which to judge hires. The whole thing is a little mind-boggling.
Having gone through the new grad recruiting process this past year and getting multiple FAANG-level offers, most of what you say is true: I did good work in the past which led to one of my offers, attending a top school certainly helped getting interviews, I built a startup outside of school which gave me ammunition for interviews and I also relied on sheer numbers to get interviews.
I ended up picking a lesser known pre-IPO scale-up over Meta and my other offers as I got matched to a top AI team at the scale-up whereas I’d have to gamble with team matching at Meta. My main concern though is that by selecting a lesser known company, I might limit my future exit opportunities even though I work on cutting edge tech in a hot field. Throughout the post you do mention about brand names contributing to your success which confirms my concern.
When choosing a new job, how do you balance relevant work vs brand name when you’re trying to maximize exit opportunities ? If you were in my position, what would you have chosen? You were a top IC in a top company - was brand name really the issue when you were applying?
I don't have profound answers to this because you've asked very hard (and important) questions. But I'll throw a few things out there:
- Though it's important to consider how your current job might lead to your next job, one shouldn't live one's entire life as a giant Stanford Marshmallow Experiment. There's a strong argument to be made for taking the job that excites you most, regardless of future outcomes. "What we do with our days is, of course, what we do with our lives." — Annie Dillard.
- That said, I increasingly feel it's difficult to get noticed without strong signals... and one very strong signal is the name of the company you last worked for. The year prior to me joining Microsoft, its policy was not to recruit at my middling school (University of Maryland) _at all_. It didn't matter how stellar a student you were, or what amazing work you did at the university. The highest order bit was literally the name of the school you attended. This tendency has, if anything, _increased_ in industry over the years as people flooded what once were considered lucrative jobs.
Of course, you could also get lucky. Perhaps you joined Dropbox before anyone had heard of it, and by the time you wanted to change jobs, it was a household name. But the experience of most people who join startups is that the startup goes nowhere. You leave without anyone having any idea how good or not you are, because there's no frame of reference. Colleges in the US used to solve this by having SATs — who's to say how good an "A" in your no-name high school actually is? But the software industry has no metric like this, so your best first-order proxy, unless you're willing to screen 2 orders of magnitude more people to get your next hire, is to filter by recognizable names.
This was an aptly timed read as I prepare for interviews to steer my career towards AI. It came across as a reassuring checklist, and I am glad that I have experienced those personal milestones in the past. Two things I do differently are:
1. Taking a more methodical approach to system design interviews via HelloInterview
Thanks for sharing Sumit, I'm curious about your 2nd point, did you find mock interviews not as useful? (I'm currently exploring a few mock interview sites, they're not cheap)
Love the last part on "Being the number one movie at the box office". Such is true in so many areas in life.
Man, what a great read! As someone currently in the recruiting process with the AI labs, I found myself relating to so much of this. My favorite part was your point about “preparing like an Olympian.” People around me sometimes think it’s insane to dedicate every spare hour (weekends, holidays, and evenings outside your 9–5) to recruiting in the weeks leading up to interviews. But in the end, those few hours can completely change your life trajectory, so why not give it everything you’ve got? Thanks so much for writing this up. It honestly made me feel way better about how I’ve been approaching things.
Awesome to hear. Fingers crossed for you!
Love the post!
I suffer from severe confidence issues and was always doubting myself after not getting screened, or flunking an interview. Have been following since your "Work-Life Balance slows careers", and your honest insights have been invaluable. I think it's rare for high achievers to acknowledge how hard they work as well, since it's nice to be called a genius who doesn't ever have to make sacrifices.
Cheers!
Thanks for this. I recently started the intense prep for system design interviews, so this strategy resonates with me. Sadly, it took bombing a few interviews to convince me it’s time to treat interview prep like a full-time job. It’s nice to feel less alone in this!
Thank you for writing this Phillip
I'm the opposite. I've given so many system design interviews that they are pretty much second nature at this point. Leetcode type problems still trip me up though.
So much wisdom, objectivity, positivity and humility in one post! Thanks for writing about your experiences!
Great article, Philip Su! Your preparation for the interviews inspired me.
A lot of valuable knowledge condensed in a post! Loved it.
Informative and thought provoking as usual!
A bit of a tangent but I had a similar experience with system design interviews (even though I've actually done a bunch of them as an interviewer myself). What does it say about the quality of the interviewing process when one who can admit a glaring weakness in this space and still apparently do a good enough job to be hired with just two weeks of intense preparation (and some luck in problem/interviewer selection) lol. I have some thoughts on this of course but I would be curious what you think on the topic.
Absolutely — your point has profound implications on the whole process of interviewing itself, the premise of it. For instance:
- What does it mean when OpenAI turns down referrals I made where I am _positive_ the referral would have been a stellar addition to the team? I've managed hundreds of engineers and am unlikely to be wrong about this.
- Microsoft insists on interviewing _internal_ candidates when moving between teams. I had many years of 4.5 reviews (a very high rating), but was still declined to join other teams due to 3 hours' interviews.
I agree completely that we aren't using much higher signal mechanisms by which to judge hires. The whole thing is a little mind-boggling.
Having gone through the new grad recruiting process this past year and getting multiple FAANG-level offers, most of what you say is true: I did good work in the past which led to one of my offers, attending a top school certainly helped getting interviews, I built a startup outside of school which gave me ammunition for interviews and I also relied on sheer numbers to get interviews.
I ended up picking a lesser known pre-IPO scale-up over Meta and my other offers as I got matched to a top AI team at the scale-up whereas I’d have to gamble with team matching at Meta. My main concern though is that by selecting a lesser known company, I might limit my future exit opportunities even though I work on cutting edge tech in a hot field. Throughout the post you do mention about brand names contributing to your success which confirms my concern.
When choosing a new job, how do you balance relevant work vs brand name when you’re trying to maximize exit opportunities ? If you were in my position, what would you have chosen? You were a top IC in a top company - was brand name really the issue when you were applying?
I don't have profound answers to this because you've asked very hard (and important) questions. But I'll throw a few things out there:
- Though it's important to consider how your current job might lead to your next job, one shouldn't live one's entire life as a giant Stanford Marshmallow Experiment. There's a strong argument to be made for taking the job that excites you most, regardless of future outcomes. "What we do with our days is, of course, what we do with our lives." — Annie Dillard.
- That said, I increasingly feel it's difficult to get noticed without strong signals... and one very strong signal is the name of the company you last worked for. The year prior to me joining Microsoft, its policy was not to recruit at my middling school (University of Maryland) _at all_. It didn't matter how stellar a student you were, or what amazing work you did at the university. The highest order bit was literally the name of the school you attended. This tendency has, if anything, _increased_ in industry over the years as people flooded what once were considered lucrative jobs.
Of course, you could also get lucky. Perhaps you joined Dropbox before anyone had heard of it, and by the time you wanted to change jobs, it was a household name. But the experience of most people who join startups is that the startup goes nowhere. You leave without anyone having any idea how good or not you are, because there's no frame of reference. Colleges in the US used to solve this by having SATs — who's to say how good an "A" in your no-name high school actually is? But the software industry has no metric like this, so your best first-order proxy, unless you're willing to screen 2 orders of magnitude more people to get your next hire, is to filter by recognizable names.
This was an aptly timed read as I prepare for interviews to steer my career towards AI. It came across as a reassuring checklist, and I am glad that I have experienced those personal milestones in the past. Two things I do differently are:
1. Taking a more methodical approach to system design interviews via HelloInterview
2. Being frugal and skipping formation.dev
Thanks for sharing Sumit, I'm curious about your 2nd point, did you find mock interviews not as useful? (I'm currently exploring a few mock interview sites, they're not cheap)