Why not? People can't fake their way through a deeply technical, probing, 2-hour conversation.
You'd be amazed just how much you can learn about someone's actual skills and experience (or lack thereof) through long-form discussion. I think we don't truly talk enough in our currently broken interview process.
Funny enough, I got into my one only and hopefully last BigTech company without a single coding interview even though my job description required me to know how to code. It was all behavioral. It was for a cloud application architect position at AWS ProServe (yes direct hire with the standard 4 year structure between base + bonus + RSUs).
My current job was also behavioral where I am a staff architect at a 3rd party company and it does require coding. As an interviewer, I also only do behavioral interviews. But let’s be realistic, it doesn’t take much to be a competent enterprise dev or even an enterprise architect.
The type of hard problems that BigTech has to solve is completely different. While I would never have trusted any developer I ever met at AWS within 100 feet of a customer, they also shouldn’t let me within 100 feet of the code that runs any of the AWS services.
Even at my medium size consulting company we have a 0.4% application/offer rate. Can you imagine what it is at BigTech? How do you filter just by talking to someone?
It’s about 10% acceptance rate once you get the interview. I don’t know what stage of interview - HR, technical or behavioral/architecture interview. I’m in the interview pool for the last level.
Now imagine there are 1000 people who are capable of submitting an application that appears to match the job description. Do you have a way to help either winnow out the 750 worst or (better) identify the 50 best of the lot to start to engage in these 2-hour deeply technical discussions?
Hiring developers is half lottery, half dark art. Best guy I ever hired was when I was tech lead for a large streamer. I hired a guy at essentially minimum wage to write some very basic HTML pages. Within weeks he was writing code. Within a couple of years he was a much better dev than I'll ever be.
I'd almost be down by literally hiring devs by picking resumes out of a hat and just having them on probation. The sheer amount of time and energy wasted having good devs doing interviews instead of doing code is horrible.
> I hired a guy at essentially minimum wage to write some very basic HTML pages. Within weeks he was writing code. Within a couple of years he was a much better dev than I'll ever be.
This sounds like a wild story, and I believe it. I love an underdog. Did you ever blog about this? It sounds interesting to read about.
No, never blogged about it. Don't want to put the guy on blast. We formed our own consultancy at one point and he ended up at a very senior position at a well-known SaaS, so good on him :)
It becomes less of a dark art, if you take a look at the applicant's portfolio. What side projects have they built? What interests them? What technologies did they use in those side projects?
There is something like that for jobs in design and art sector. Checking what kind of projects someone already did. Checking their website. How does that look? Is it well designed? What are they showing on their website about what they have done?
But the issue with these ideas is, that apparently these days there is zero time planned for actually checking candidate background properly like that. That's why hiring a dev can feel like a lottery. Hiring people work with utterly incomplete knowledge.
The leetcode style tasks have nothing related to coding for the past 2 decades. That part is solved for a long time now. They ask for knowledge, which is a search away from everybody. I don’t know anybody, who knows those, and not only because of interviews. Also interviewers ask these, yet average code isn’t optimized at all. A simple question, like what’s your opinion of <anything> will tell you more than any leetcode question.
What happened with recommendation letters? Portfolio?
If I have 10+ years of proven experience with positive public recommendations written by high managers from F500 companies why should I be challenged with puzzles?
Talk an interview to figure out motivation and if we have the same vibe. Then hire me and decide a month later if I'm worth your money.
All of these are easily faked and/or not able to be shared. You think I can share code from FAANG?
Also, hiring is very expensive for those first few months at big tech. It’s also very morale draining to see your peers fired constantly - which is what would happen. Morale at faang is already hitting rock bottom due to the mandatory yearly+ 10%+ layoffs.
Interviews tuned properly for the position are almost always 95% talking with a decent trial/probation period. If they have the general credentials and can talk competently about the detail work, there is no benefit for either party doing these asinine code challenges.
The problems that spawn these leetcode barriers are WAY better solved at the source.
Prob 1: the interviewer doesn’t know shit about the actual role. I know it’s an uncommon opinion, but this should just not happen ever. I would gladly sit in on a speaking interview for my would-be coworkers, with the hiring manager, to check this. And the actual manager should be present and knowledgeable too. But 3 employees doing something is too much $ I guess.
Prob 2: credentials and high degrees are prolific, super mandatory, but also treated like they are meaningless. Experience is borderline ignored entirely. Worked with a software deeply for 10 years, but your cert is out of date? You won’t even get past the computerized HR filter. But because this is so openly stupid, companies tack on leetcode as another filter instead of rethinking the value of experience and relying on probation and bullshit detectors.
Excellent points. I just wonder about the 2-on-1 interviews. I think they tend to double-down on the one-sidedness of typical interviews.
It's not a group engineering meeting. Unless it's big-tech hiring of fungible worker drones, everyone in that meeting should be trying to get more signal about the others as individuals.
I've been the "1" in 2-on-1 three times recently, all for startups. For me, it's hard to think about the hard-thinking questions, and also be finding rapport and getting signal from two different people at once. I'd much rather spend twice as much of my time, to get each person 1-on-1, no additional cost to them.
Two of the recent 2-on-1 interviews were on videoconf, and, with the videoconf setup they chose, I couldn't even see one of the people most of the time.
The other interview was 3 people in tiny conference room/closet (the size of what used to be a one-person cubicle), for an hour, so it was especially stuffy and crowded with 3 people, and a great way to get Covid.
Besides more people complicating the situation, people tend to speak candidly with me 1-on-1, but are less likely to do that if their colleague is in the room or on the call.
Here's an intuition: Imagine you were a company interviewing startup candidates, and HR suggested saving time by you interviewing two candidates at once. It would be awful, and you wouldn't be able to get much read off of either (other than to tell if one or both were sharp-elbowed). So why do it 2-on-1 in the other direction, unless you're saying it's not important for a candidate to get signal about the company and colleagues?
Oh, I actually turned down a mostly great startup opportunity, basically over an n-on-1 interview.
I had a strong recommendation (halfway to hired, before they even met me, with a colleague's favorite ex-manager/mentor) to take over engineering and technical leadership, from a startup's technical CEO.
After I passed his interview, he had me do a 3-on-1 interview with the 3 engineering team members.
They all seemed a little awkward with the circumstances or format, and none opening up.
I passed the 3-on-1 from the CEO's perspective, and he gave me a written offer.
After I confirmed with him that one of the 3 existing engineers didn't want the role (I'd gotten a little leadership role vibe from him, despite 3-on-1), I asked to meet with at least one of the engineers 1-on-1, so I could get a better feel for the team.
The CEO pushed back hard on that, because what you see with them, is what you get.
We don't all get our intuitions the same way. Maybe the CEO could've gotten all the info he needed in a 3-on-1 videoconf meet&greet, including all the info I would get from 1-on-1. Or maybe I could've gotten info that he couldn't have, yet he seemed to be rejecting that possibility.
It might've been a great situation, but I ended up not taking it, arguably traceable to a low-info n-on-1 interview.
I have been running the exact same pair programming interview for ~100 candidates so far. Make a little exercise that is representative of your work, run everyone through it. Whatever you test for, that's what you should expect to get.
I mean, if the current Kafkaesque nightmare isn't working, then yes? You might as well read chicken entrails, at this point. At least then you'd get some diversity, if nothing else.
But the actual solution is to not let bean counters and bootcampers do the hiring, but we're a few too many programmer-generations away from anyone who actually knows what they're doing being left at most of these companies, so.
