After the week was up I went into the onsite and in the "technical" portion of my interview two engineers went over the code I had written for the assignment and asked me about design choices I had made and what I would do if constraint X was added or feature Y. It was all very open ended and much more of a true discussion than an interview which I really appreciated.
I think this kind of format is ideal for interviews. The assignment requirements were simple enough that you could fulfill them easily and in your most comfortable language without any time pressure, but you could also go above and beyond and show that you really knew your stuff. For example, in the instructions they didn't explicitly ask for error handling on the input, but both of the engineers I interviewed with really liked that I had included it. You weren't penalized if you did just the bare minimum because you had the opportunity extend and build on the assignment during the onsite. I felt enabled to showcase my knowledge and justify my design decisions and that that effort was rewarded.
They then come back for a code review, walk the team through their code and explain it. This tells us a lot about their ability, but also their communication style and clarity, how they approach problem solving in general and even in which problem they chose.
I chose this approach because 'coding under duress' during an interview does not offer a valid sample of what the candidate can do. Taking the time they need and working in a comfortable environment will give a much more realistic result. Coupling the take-home with an in-person code review weeds out those who may just Google up or borrow code from elsewhere.
Also, I highly suspect companies who are having trouble hiring senior devs are having issues stemming from not paying enough for their given situation.
Companies like Square and Stripe have project-centered rounds during their onsite that allows them to assess these skills without giving a take home.
In my case, I do a lot of writing these days and, if someone were to ask me to write a thousand word analysis on some topic I was familiar with, I certainly could knock that off in a few hours. But, assuming I agreed to do it at all as part of the interview process--I'd be far more inclined to just give them a bunch of links to my work--I'm going to put the effort in to come up with a tight, polished product.
I'm not going to jump through stupid hoops, but I love jumping through interesting hoops.
1. Require developers to code on demand during an interview (personally, I have a hard time with this; the pressure of an interview is not a normal working situation).
2. Allow someone to do it on their own time, under no time pressure. (I advocate paying them for their time at a fair market rate).
Do you see another way of a senior developer demonstrating their abilities without hitting either of the two above?
For the junior folks, sure: have them burn through homework.
Aren't you already spending many hours researching a job as the part of the process? Aren't you already taking a full day off work for the interviews? Why is that time all fine to spend, but not fine to spend a couple of hours on doing an assignment?
I've never had a full day interview for a company I'd actually want to work for (based on what I've learned while grinding through the interviews) so I'm beginning to take requests for those (along with requests for advance coding exercises) as a recruiting smell.
My team has full-day interviews and I'm curious to see who has been self-selecting out of this.
We have had a perpetual shortage of developers though, it sounds much harder to land a job in the US.
But even when the interview process was mostly a formality because I knew many of the people and they knew my work, I still had some on-site interviews.
I'm pretty sure on-site interviews are the norm for the vast majority of professional jobs. Including, or perhaps especially, for the most senior positions.
I'd much rather do a take-home test than spend a full day getting grilled in interviews.
Anyone with any level of self-confidence, worth and respect.
I suppose there are distributed companies that just do things over video link these days (and I've interviewed people over a video link when necessitated by travel schedules or people being in different locations). But some number of in-person interviews over the course of 4 or 5 hours is absolutely standard in my experience.
If I'm unemployed, then full day is fine.
If I'm really keen to get out of my current job, or really keen to work for you, then I might do it.
But if I'm just exploring opportunities, or I have other options in progress, then I'm not taking a day off just to interview.
With any interview when it might be really obvious 10 minutes in that it's not going to work. If I've scheduled that during my lunch break, or for an hour before/after work, then that's a small cost.
If I take a day off work to interview, then that's costing me in the order of $1000. If I don't know whether I want to work for you, then why would I do it?
And I definitely can't do that for 5-10 different roles that I might apply for.
A full day is also quite hard work. Interviews are stressful. Dealing with people you don't know, trying to make sure you don't do/say something stupid, it gets exhausting.
I generally expect 4-6 hours worth of interviews before an offer, but the typical process stretches those over a few weeks, which allows the candidate to fit them into available blocks of free time, and gives the candidate oppotunities to think about how things went, what questions to ask, whether this seems like the right fit, and pull out at any point.
For permanent jobs, it's usually an hour long interview followed by a programming assessment; either take-home and then presenting to their developers, or codility. But for freelance work, none of that. Just a talk and they hire me. But of course if I turn out to be useless, a freelancer is easy to fire.
(I don't like video interviews much.)
As for hiring freelancers, I'm not involved in hiring programmers but we use external people for various other things. We've try to use people we have experience with but it's mostly not a big deal. If we don't like the work they do, maybe we're out a few bucks but we just don't hire them again.
Coding tests are just an early stage filter for experienced people - so those companies get dropped early on.
You don't want to get your developers together for any random loser who'd never want in your team anyway. So the coding is a late stage filter.
At least for me it's always been.
And Martin any professional should adapt to the systems in place - and I wouldn't use terms like "random luser" might have some blowback.
And if they're not interested in putting any time or effort into a new job, it seems to me they're probably not that interested in that new job in the first place.
But it seems weird to me that so many people here consider a whole day of interviews to be totally acceptable, while some actual coding is not. Compared to some of the interview horror stories I often read here, this seems vastly preferable. I actually get to show what I can do in a realistic setting, and I get to explain why I do what I do. I get to meet the other developers and talk to them.
And with the best programming tests, I actually get to learn something new. For the best one, I had to learn React (which I had no experience in), build a game board on which you can put obstacles, and write an algorithm to find the shortest part through the resulting maze. (I ended up rejecting that job because it was too far away and not enough pay, but I'm still really happy I did that coding challenge.)
I do agree about a daylong set of interviews with 4 or 5 people that's also a filter id use.
In my experience, these sort of tests do often seem to be used by the more interesting, more competent kind of companies. The first company where I encountered this (14 years ago), had only highly competent developers, many of them committers to various open source projects. About the recent one where I wrote a customised version of A* (because of the presence of wormholes, which made it not remotely rote for me or for most people, I'd expect), I of course don't know how it would have turned out had I taken the job, but if they select for this kind of thinking, I see that as a very positive signal about the kind of work they do and the kind of people they hire.
At companies that didn't have this kind of test, the quality of their developers was a lot more hit-and-miss.
At least for permanent employees; I've never encountered this for freelancers.
I certainly didn't get asked coding Questions when I went to BHRA (ranked 1 or 2 in its field) research organisation.
I was told in my first week "oh there is a book on Fortran in the library" teach yourself it
But it's very likely different for research institutes. I'm talking about companies.
I should clarify that I was thinking more of outfits that respond to my enquiry with "Hi $candidate, please do this generic exercise after which we'll deign to look at your CV". Bonus points awarded/deducted if the requested exercise is for something already available on my github profile.
And 4-6 hours is typically my free time (not free computer time, total free time) for a week.
In practice there’s a pervasive phenomenon called asymmetry of effort. The hiring manager may crib a task set either from a perfunctory google search or their own body part. The sum total of time spent on their part is often five minutes, including coming up with the problem and their review of your code.
This follows no industry standard, and I can count multiple experiences where they were too inept to “grade” the homework assignment, thinking it didn’t work when it did, or that the problem set was impossible.
The reason senior engineers in particular turn down homeworks is because they’ve been burned by them before, and the average hiring engineer tends to know less than they do (but as the hiring manager, can’t assess it, mistaking competence for arrogance or ineptitude).
When a company is not willing to invest the same amount of time to interview as they ask of the candidate, it’s a strong signal of how serious they are as a quality employer.
When job hunting as a senior, hearing about take homes (esp. early stage ones) is the young person’s dating equivalent of considering a relationship with someone divorced twice and a felony. You just get conditioned to pass up kissing frogs.
At my current employer, our recruiting manager has a preliminary talk with the candidate and if there seems to be mutual interest in continuing in the process, she sends a simple task. The task was defined by the technical team, it should be short (no more than an hour) and it has nothing to do with our work.
Once she receives the submitted task, we ask everyone in the tech team to make a review. We do have a "gold standard implementation", but we know that no one will get it and we don't care. We had cases of candidates that misunderstood the task and we had no issue with giving them a second chance or simply asking what was unclear and what would they do differently had they had got the intended requirements.
We just want to filter the obvious "no hires" from the ones with potential. The recruiting manager communicates this even before the task is submitted to them. I am yet to hear any story of any candidate that was interested in the position but walked away due to the way that the process is conducted.
Thank you for making my point. Why would you ever hear about it?
This is the "British Rails" fallacy in action. (A long time ago, the British Rail organization had heard complaints that residents of a local community weren't satisfied with the once-daily train service it received. Its response was there was no need for a second train, because at the times considered, no one was standing on the platform waiting to board it).
Nice coconut earmuffs.
Seems like the one that does not want to listen is you.
The take home task is only communicated during the preliminary interview with the recruiter (i.e, it is not something that is required to start the application process) and we do get feedback from candidates that have withdrawn their application. And while we had cases of people that withdrawn for various reasons (company size/culture, process adopted, being too "startupy" for a German company, or even simply getting a better offer somewhere else) the take-home task is yet to be mentioned as an issue.
In the end, your reasons for being against take-home tasks are not really against the idea of take-home tasks, just some of the implementations you've been through. Your "point" does not stand.
Everything else is simply busy work for all parties because it doesn’t tell you anything you don’t already know, plus it scares away quality candidates.
You don't even really need a probationary period in an "at will" state.
At some point, you either trust the credentials from universities and certification bodies, and take what the resume says at face value, or you have to try to reinvent some wheels, and re-discover that eliminating the phonies and mimics is a very difficult problem that even the experts can't solve entirely.
I don't know what the conversion factor is between running a fruitless job search filled with cargo-culted bozo tasks, and just randomly picking the first vaguely suitable candidate off the stack and letting them run for a few months, but I suspect that filling the position quickly and painlessly is probably worth about one month of having a 0%ile employee, or six months of a 20%ile, or keeping a 40%ile employee until they retire. This rests on the presumption that some work done right now is worth more than the same amount done some time in the future, and that higher %ile-ranked candidates will be less tolerant of any stupid hiring games.
Does anyone know if having different tracks is legal?
So essentially about 4 hours of time. What makes this different from a company that wants 4 interviews from different managers, which likely takes up more time plus all of it being fixed time slots, versus having one hour of work that you can work on at your own pace?
It means you don't do unnecessary work, you avoid pitfalls and traps you have seen before, you don't over-engineer but you keep it as simple as possible. It also means that you push back against non-constructive requests from business and management and focus your time and effort on what matters.
I also avoid anyone who wants me to do a coding exercise for a few hours over the weekend.
If you want pay me for that time, and make it not a dig and fill hole exercise - find a OSS project that needs some contributions. Decide that you and your team will do thsoe via your ongoing interview process that hires people before the need arises, so you can all take your time selecting and onboarding people.
Value your own time, or no one will
I could set up one of our OSS projects with first-contribution issues and help walk someone through but I suspect most people would rather not spend the time. If it takes 4 hours with back-and-forth in the PR (simply because I may be asleep when you commit and you when I comment) that's a big request of any person.
It should also mean you don't under engineer, but that is extremely hard to test for because that involves knowing the business and your customers and having a reasonable estimate of where future needs will be. I'm not sure how you would test for lack of under engineering, especially since any interview task would be a perfect case where you practically can't under engineer since it is guaranteed throw away work. Maybe asking during code review for how you would've done the solution different if you knew that in the next quarter you would likely have to implement either feature A and B or C and D (but you didn't know which yet).
The main antagonism here is that the hiring company wants to minimise its effort in getting great devs, and the devs want to minimise their effort in getting great jobs.
The first point to note is that constant marketing is the first, best solution to this. I would definitely put more effort into joining a well-marketed company (StackOverflow?) than J.Random Inc. So both me and J.Random Inc had better put some effort into standing out from the crowd.
OSS is one seemingly good way to do this - and having a reasonable Github account is something I would say should get you through most interview filters (ie if you are looking for a Python dev, and they have commits to say a flask extension or bits of salt-stack, then you can blip over the Fizzbuzz and whiteboards.)
But marketing is simply a way to get past other people's filters. "Networking" is good, whatever that might mean, and being a good citizen works. but these are as always, long term, constant application efforts. And of course, costly.
Secondly get over the "we only hire the best of the best of the best". If that's true then like the SEALs you clearly have a six month training programme that takes the best and shapes them into effective teams, fully paid while they learn - yes? And the training staff for this programme are pulled off current fee-paying projects to keep up the standards yes?
Otherwise your best-of-the-best is so much auto-trumpet blowing. So please leave that at home. Focus on process not heroes.
The interview by takehome project is a big problem.
Yes it clearly weeds out candidates - mostly by making the ones with options elsewhere go elsewhere. (Even Google, with its firehose of applicants, seemed to realise this was a dumb idea and instead used the MIT graduate program as a filter instead).
My main issue with interviews-by-homework is it is usually a hole-dig-and-fill session as the problem has been done by dozens of people before (often you can find their solutions on github). My time is then valued at less than zero. As a filter before we get to interview, it really is just an artifical hurdle. All you are saying is that "we have soooo much choice we can make you dance before talking to you" - if it works for you great. Your marketing is working (see above)
IMO a better approach is to find different existing properties that you can filter to get applicants to the interview.
Sometimes these are "Alpha Male" filters like 1st class honours at MIT. Other people on HN stand by different filters - I seem to remember someone saying one of their best hires had taught themselves coding whilst getting off drugs in a Glasgow slum. How you filter for that is harder than the MIT problem but I bet its a less tapped seam than MIT.
Overall, a good interview process is probably focusing on the wrong end of the problem. If you have a new open position and then you start looking its too late. Be a good OSS citizen, be involved in the community of developers (OSS and elsewhere), find ways to reach out to unusual developers, or people not marketing themselves, and keep the process going constantly, with suitable investment. The general Big Co idea of "you cannot put out job ads before you have a signed off budget and position" tends to make this a problem. A solution would be to take the expected Agency percentages for each new hire (between 5-20%) and put it into a centralised "developer evangelist" team that just gets people through the door.
This way when you need to open a new position, you will have four people in mind you already want to hire, and you can just go to Starbucks for an interview.
But coming fresh and new into a company, it's hard to just have your opinion or taste taken seriously - especially where that opinion will be unpopular (ie your code base needs rewriting from scratch)
So FU money is needed to avoid the tendency "just to go along with it for a bit"
Hire rich people is hardly a great line, but it will have benefits.
Hiring desperate people with no other options does not sound like you are winning of course
I rarely start anything these days without thinking about the problem thoroughly. It results in longer development times, but better designed, and more robust software, or so I hope.
What if 9 great companies give just 2 hours of homework and 1 gives none?
I don't do challenges any more, life is too short to waste solving problems that don't exist for someone else without getting paid.
The problem it's trying to solve is just the surprising number of bad candidates that apply. We're talking about people with decent CVs, a bit of work experience at different companies, some open-source work etc., and who had a good chat on the phone – sometimes they just completely mess it up when faced with a task like "combine these JSON files to make output that looks like this sample". Things like code that literally doesn't work, or code that uses totally inappropriate tools. We want to avoid wasting both their time and our time by bringing them onsite for an interview that isn't going to go anywhere.
The risk with being "hired to do real, paid work remotely" is that it relies on finding people who are in a position to do that, and this ends up excluding a bunch of good candidates. Not everybody is in an employment situation that will allow them to do external work, and it disincentives people with families or other commitments.
The process of applying for a position shouldn't be over-burdening on a candidate. I agree that it's unfair to expect lots of stupid work, multiple-day interview processes and so on. But I don't think it's unreasonable to expect that a candidate who wants to work with a company can afford to spend a few hours in total as part of that process.
That excludes anyone with a no moonlighting agreement in their current contract, though.
A simple 100 line code can be curated and taught by a friend/relative in a very elaborate way. This approach can negate that issue in someway.
I generally think this works well for a few reasons:
1. The take-home task lets us see if candidates meet a minimum standard of competency and generally follows good practice – do they test the solution, document it, write clear… of course, I think it's important that the size of this task is limited to avoid wasting candidates' time.
2. The on-site interview helps to make sure the candidate is able to discuss the work they're doing, collaborate with other technical staff, and communicate technical ideas well.
It feels like the overall goal of an interview process is to ensure that the candidate is at the skill level they have communicated, and that they can work with others.
You've got me thinking that maybe I need something a little more complicated that will give room for those questions.
A good template is:
- Simple yet relevant take home coding test with a relaxed timeframe
- Technical portion of the interview where the candidate is asked to add a simple feature to their existing tech test.
- This is followed by a more general interview.
None of these sections should take more than 1-2 hours.
I like hackerrank/codility style coding challenges.
They're timed. They're run against a standardized test suite. And they have support for many languages.
It's really unfortunate the number of companies that give custom take home coding challenges that run the gamut. The problems with which are manifold:
- Inconsistent technology and language choices by candidates
- Time spent varies wildly and is easily underestimated by companies - and 4-5 hours for an unpaid coding challenge in an early round with a company is a significant investment
- Good candidates will de-prioritize your company against others that move faster
I know that I only answered a small subset of the question, but it's very difficult to lump all developers together and come up with a single, wholistic, and ideal interview process. From the nature of the role (IC/Lead/Manager) to the technologies/specialty to the company's mission, the process will need to be customized to some degree.
However, do not think that your custom coding challenge will somehow be the perfect key to assessing developers.
The advantages to that approach, as I see it, are:
- the developer is tested for knowledge that is most relevant for their current job (as opposed to writing general-purpose algorithms)
- the developer is showing their familiarity with a testing environment
- writing working code in a code editor seems to me more practical than writing (pseudo)code on paper or on a whiteboard
There are, of course, disadvantages to this approach as well. The developer may have hard time working on an unfamiliar machine and not in their favorite code editor (though what we had was quite standard). This might be a tad stressful (although I don't know how this compares with solving a general coding problem on a whiteboard, or with a task asking you to think not just of a way to solve a problem, but of a performant way too). And it made us focus on the candidate's coding skills too much (which some hiring managers might find objectionable).
* It's way too easy to set a project that takes too long. I was given one the other day that would have taken about two to three full days. That simply doesn't happen if you do the test sitting in front of them - the time spent (and potentially wasted) is always capped because the test setter doesn't want to waste too much of their own time.
* Take home projects inevitably have issues ("bugs", if you will). You can't just "resolve" those problems if the person who set it isn't sitting in front of you.
* Take home projects take employer skin out of the game. It's way too easy to give out twenty tests and ignore ten of them. I was actually half way through a homework test the other day when I was told that surprise they had an open offer and the candidate had just taken the job. I doubt they'd have wasted their own time testing me IRL knowing what they knew.
From the employer's perspective:
* In the last case I also realized after googling that the answers were online and I could just have copied someone else's.
* You lose a valuable source of feedback which allows you to iteratively improve the test if you're not watching the candidate while they do the test. I suspect this is partly why the take home tests I've been given are generally of a lower quality than the in person.
* It's valuable to talk through the solution with the candidate while they do it (describing trade offs they made, pitfalls they're cognizant may crop up, potential downsides to their approach).
> It's way too easy to set a project that takes too long.
> Take home projects inevitably have issues ("bugs", if you will). You can't just "resolve" those problems if the person who set it isn't sitting in front of you.
> Take home projects take employer skin out of the game. It's way too easy to give out twenty tests and ignore ten of them
This gives the employer skin in the game: they're not going to keep their developers off their work for dozens of candidates who are probably not going to make it; they only bring them out to assess a very likely hire.
I guess some employers use these coding assignments badly. You're totally right to avoid those.
> I could just have copied someone else's.
> You lose a valuable source of feedback which allows you to iteratively improve the test if you're not watching the candidate while they do the test.
> It's valuable to talk through the solution with the candidate while they do it
I don't believe someone who can't code can explain someone else's code in a convincing way.
I get the advantages, I guess everybody does, but there are disadvantages too.
I'm also fine with take-home challenges, though. But you're absolutely right that they're often vastly underestimated, and you shouldn't give someone a take-home assignment when there's only a small change they'll get hired. It should only be done once you're sure you want to hire this candidate if they're as good as they seem to be. A take-home assignment is too big to sift the chaff from the wheat in a large group of candidates. But once you're sure you'll hire them if they do well on it, a take-home test is totally fine.
Though such a take-home programming assignment should also involve presenting the result to a group of developers to explain their choices. If you're not willing to commit to that, you can't expect them to commit to putting in the time to do the assignment.
I refuse to do other types of take home challenges, but am fine with these.
Coding: I actually like coding tests in interviews. Then I know my potential future colleagues will also have had a coding test. The company cares at least a little that my colleagues can code.
Just don't take the coding test too seriously. Don't expect me to write correct syntax on a whiteboard, for example. It should be more about having the right ideas, or even be able to discuss possible approaches if a solution isn't obvious immediately.
Also, make it a little challenging - no computing of Fibonacci numbers or FizzBuzz.
I liked one where I was asked to implement Quicksort (not that original, but OK), but they also gave me a definition of Quicksort along with it. That seemed fair.
This is important. I'm nervous. I don't know you people. And modern IDEs/Editors fill in most of the boilerplate for functions/classes, etc as well as help me recall library function names.
Be respectful of my time and don't expect me to practice whiteboard coding for a month just to interview with you.
As developers we are a bit introverted so being in a room with 2-3 people knowing that after this there will be another 2-3 people for round 2, is very difficult. As an interviewer I’ve watched candidates open up talking about what they know and why they enjoy being a developer and completely fall apart (sweat, shake, refuse to get out of their chair) when you hand them a dry erase marker and show them the white board.
It’s really that simple.
If 3+ professionals are willing to vouch for me and I have 20 verified years working at major tech companies, do you really think I’m a guy faking I know how to code?
I wish it was as simple as checking your code and references, but I think it is not.
- how you'll interact with the team
- whether our company culture is something that you'll enjoy
- whether my team will become a better team when you join us
- maybe you're defensive about your code and do not like having it reviewed
- maybe you enjoy pair programming
As a hiring manager, how do you find answers to these questions?
Here are a few things I do that help me better predict whether candidate will fit the team:
- ask candidate to do coding task at home and later discuss the solution during face-to-face interview. While discussing the solution and suggesting improvements I can see how candidate reacts to feedback, whether they can explain why they came up wit this particular implementation, what do they say when I suggest an idea for refactoring etc.
- I like having another team member with me during interview. That person is an observer and they help me review candidate's performance (e.g. it happened that I had an impression that candidate did quite well, but my observer told me that I was often helping candidate with their task, which I had not noticed)
- lunch with candidate might be helpful. I take 2 team members with me and 4 of us grab some food out of the office. This helps candidate meet potential teammates and talk in a more relaxing atmosphere, so both sides can see how they feel about each other.
- one more thing I do is during initial phone call when I tell candidate about company, at first I give a very brief introduction, then before I say more, I ask whether candidate has any questions. This gives me a lot of insight: some candidates ask advanced questions, some ask very basic ("so what does your company do?" happened a few times!), some focus on product, some ask whether they'll be able to learn a particular skill. Some candidates don't have any questions, which is usually a red flag. These questions themselves are not enough to decide that candidate won't fit, but they're helpful.
- I ask candidates what they would like in their new job to be different than in the previous one. One candidate tells me they want to focus on one particular JS framework. If in my company we chose libraries and languages on per-project basis, she might not like it. Counter example: candidate that wants to learn multiple new frameworks might not feel well in a place that is conservative when it comes to introducing new libraries
And if people do well with pair programming, let them do it!
We love to think that getting a job should be based on pure meritocracy, but being able to do the job is one small piece of the puzzle.
It's the personality, cultural fit, and general interpersonal skills that really make or break a good team. If you have a team of the 5 best programmers in the world but they can't handle basic human interaction, nothing will get done.
No BS interviews, people actually believe you when you say you can code.
The best fulltime jobs I had had interviews in the same way. Quick meeting to ensure that I am not a psychopath and to talk about salary. That's it. No fizzbuzz, no inverting trees. When can you start and let's get coding.
- Ask me questions that actually apply to the job. If I'm building an iPhone app, please don't ask me to run through a gauntlet of white-boarding/coding challenges that don't apply to the job. Have me walk through something I've built or talk about how I would build a theoretical iPhone app. Most of us aren't building software for self driving cars so please quit acting like we are.
- If I'm writing code or solving problems, give me an ideal scenario to do my best. Let me use my own computer (not one that you just handed me setup to your preferences.) Potentially let me do it at home (where I am comfortable and not in an unfamiliar place.)
- Gauge confidence on the technical stack
- Don't say we'll be in touch shortly and then ghost. You can say goodbye to someone without false promises. I'm likely to tell fellow developers about how the process went.
- Have people with decent social skills interview
- If I ask questions like, "What should I be prepared for?" please give me a basic agenda or guidance. It'll go much better for both of us.
Got a decent CV? Good luck trying to juggle applying for more than 2 interesting opportunities at once.
We mostly hire full stack web devs. IMO It's impossible to really test the abilities of each candidate across the changing landscape of front end, back end, DB & devops tech within an interview process that doesn't use a vast amount of time for everyone.
Instead, we don't whiteboard or code at all in our process at the moment, and try and get it all done in a face to face hour or two by:
* Taking their experience at face value. Examples: If they have been coding for a couple of years, don't waste everyone's time with fizzbuzz. Assume they will be able to adapt to our source control system, if they have been using a different one.
* Insisting on real world examples when asking competency questions.
* Asking generic questions about code, such as "What is clean code?", "what should you take into account for password security for a web app?", and looking for their ability to communicate as much as their actual answer.
* Looking for areas of strength and weakness to compare across candidates, rather than trying to catch them out.
* Scoring highly for enthusiasm, flexibility and a willingness to learn over pure technical knowledge.
I appreciate this approach wouldn't work for all organisations but we've done really well out of it.
No fancy algorithms. No obscure data structures. Just simple loops and conditionals in any language.
It has been the most effective (and most depressing tool) I’ve seen in eliminating the myriad of fakers and unqualified people.
But the main thing for me was the arrogance of the company/CTO, was pretty much a no-name e-commerce company that was expecting someone to do days worth of work for a job. I've not even heard the like FANG or anyone expecting that during their interview and they are actually places some people really want to work at. Which made me worry about the quality of developers they had if they all had such trouble in the job market that they had to spend days on a tech test to find a job.
Weird thing was, I said I was willing to be tech tested in their office over a period of hours I just wasn't willing to do a take-home tech test, apparently that was too hard for them to figure out and they just did the good ole no contact rejection.
It would show great management skills and you could save them a ton of money outsourcing.
1) can you write software?
2) can we tolerate writing software with you?
3) can you tolerate writing our software?
How would you like the interview to go so that you can decide if you want to work there?
Interestingly, I would like to see a variety of people that I might potentially work with. I'd like to pair program with them. I mean, really pair program -- with each pair writing some code. I'd like to write some code and see how they react to it. I'd like to see them write some code to see how they approach it. I'd like to work though some simple conflicts to see how they react.
I'd like to see some code (and I'm willing to sign an NDA). I'd like to work through it with some of the people who work on it and ask questions. I'd like to see how firmly they hold on to the existing code and how open they are to new ideas.
I'd like to talk to a few people in management. I'd like to see a few plans (again under NDA). I'd like to see what kinds of statistics they collect and how they think those statistics help them.
Finally, I'd like to talk to management about how they do reviews and see some statistics about attrition rate, typical pay raises per year, how many stocks actually vest before people leave (yeah, yeah, NDA).
That would be awesome.
My favorite interview process is where I’ve done a screen interview or two and then do real work to see the process as it exists. Everything else is advertisement and I’ve grown to distrust the evangelizing people do about how they work.
Most startups basically try to wing this without any definition of how they want the interview process to work. IMHO this is about 50% of the problem. On the flip side, large companies appear to be overburdened by process (anecdote example - I've heard from several people that getting into Google takes 6 months on average, along with the notorious b-tree whiteboard process).
That being said, I think Aline Lerner and Triplebyte have some good ideas on the topic:
 - http://blog.interviewing.io/
 - https://triplebyte.com/blog
I don't think implementing either one of their services is a silver bullet, but are likely n% better than what most companies are doing.
I can only speak to my relatively recent experience, but it was <5 weeks from application to offer for me. No idea if that's typical.
I don't often interview developers, but when I do, I try to get into a situation where the candidate leads us through solving a programming problem, and the issues that come up. The most satisfactory case was where the candidate and I jointly investigated an issue that neither of us had the answer to.
Unfortunately, this is not an approach that many people have much experience with, and it seems unfair to penalize people who find it disconcerting, but when it works, I think it is most satisfactory approach for all concerned. It is not easy to automate either the practicing for or execution of this style of interview, which may or may not be a disadvantage.
I've given this question a lot of thought over the last couple years as I've lead teams that have needed to get organized and expand quickly. Here's my summary taken from a Google Slides presentation I put together titled "Hiring in a Time of Cargo Culturalism".
It starts with Principles and Guidelines:
- Hiring cycles will be structured and as short as possible.
- When we start a hiring cycle, we will finish it by hiring the most qualified applicant who accepts our offer.
- Every applicant will receive a response within 48 hours and be updated on the status of their application at each step asap.
- The hiring process will be as transparent as possible.
- Objective and fair-minded measures will displace biased and bigoted ones.
- Every applicant will appreciate their experience, even the rejected ones.
- The process will be agile and adapt over time to improve and meet the specific needs of the organization.
- Onboarding will begin with hiring.
Then an outline of my team's current Methodology:
- A thoughtful and literate job posting will accurately describe the job and foreshadow the company culture.
- Simple challenges and honeypots will filter serious candidates from the applicant bots.
- At the end of every step, we will inform the applicant what comes next. Courteous templated responses will be promptly delivered.
- Two interviews. No more than three. The coding challenge will represent a genuine work sample. It will be no more than one or two hours.
- Candidates will be evaluated using a simple quantitative assessment of core competencies (see Ch. 21 of Kahneman’s Thinking Fast and Slow).
- Final decision will be a collective decision of the hiring team.
- After hiring cycle is complete, hiring team will hold a retrospective.
I've hired over a dozen developers this year. They haven't all been homeruns but no strikeouts either. A few singles or walks. A lot of solid doubles. And that's mostly what my company needs.
Also, I suspect our job description and pre-interview process (requesting a cover letter, asking a couple short-answer questions in a personal message as soon as we receive an application) succeed in filtering out candidates who aren't thinking about a job as a social inter-personal commitment.
* The session takes place at a pairing station - two keyboards/mice, two monitors (mirrored).
* We work through a fake problem that is relevant to real-world problems that we actually work on. No brain teasers, no complex algorithms. Basic software engineering using some of the tools the company uses.
* The problem is exactly the same for every candidate. This is essential; you need to be able to compare apples to apples. You can't use "real work" because real work is different every week.
* The stuff I look for is pretty mundane - can they name things sensibly? Do I have to push them or do they naturally drive out behavior with tests?
It takes 1-2 hours. It's partly a way of evaluating candidates, and partly a way of communicating "this is how we develop software at this organization". Some folks really respond well to it. Some don't. That's ok.
At the end of the pairing session. I know 100% if I want to extend an offer. I've gotten some really great hires out of this. My bad hires ended up bad for reasons unrelated to technical ability (like, their life was a trainwreck).
Everyone is so focused on finding this mythical perfect candidate, instead of giving people a chance. Which is really another way of saying "We're afraid to fire people."
Culturally, we should hire and fire more freely.
I once got a call from a company that said I was an absolute perfect fit for what they were doing-- would I consider a 6-month contract-to-hire? I had been in my job 10 years and told them no. They were so flabbergasted they called back and asked if I wouldn't consider it, that the CTH was simply to see if I was a "culture fit." I told them in plain language that if I was such a perfect candidate, they could hire me straight out. That I'd been in a job 10 years, was obviously happy, why would I jump ship so they could dangle employment as a carrot in front of me?
I can't tell you how many jobs I've not gotten because I didn't get some gotchya question-- or in some cases understood more than the person interviewing me.
The abusive hours need to end, as does the concept of "culture fit" which is just a proxy for age, race, religion, binger drinker status, etc.
We need to stop judging people and trying to feel better about ourselves by dismissing people who can't answer questions we just googled ourselves. Almost every company I've ever talked to claims they only hire the best candidates. That simply, cannot be true.
How is anyone supposed to grow if you can only get a job you're an expert in? And when what you're supposed to be an expert in changes every two years?
The answer is, culturally, you're not.
I was a formal full-time employee at NCC Group until just a month ago, when they fired me with no warning, no explanation, and no severance.
I don't want to leave a secure job for an insecure situation because they're difficult to find. They won't consider a direct hire for a "perfect candidate" -- proving my fears.
My point which perhaps I should have been made more clearly is that they're asking me to leave a direct hire situation for a temp job. They're asking me for a commitment they're unwilling to make themselves.
Power asymmetries suck.
No, you could also find this if the legal system stopped encouraging meritless suits.
Suing would be less frequent then, but as an effect, not a cause.
It was an in person pair programming session, with the interviewer driving.
Most importantly, it was collaborative, whereas many interviews often feel adversarial.
It also tested how you communicate, since you weren't the one typing.
However, it still tested your coding skills, but because the interviewer was "on your side" with respect to the desired output code, it wasn't a dealbreaker or stressful situation if you forgot the signature of an important method or function.
But it also touched on actual coding skills, while not requiring you
My ideal interview proccess thus is:
1) Talk to me for 30 minutes (not your HR, an engineer that is trained for this task). Fizzbuzz me if you have to, but preferably at this point in my career, don't.
2) One or two onsites that last a maximum of 2 hours and are early or late in the day, so that I don't have to waste a day of leave. Don't fizzbuzz me or ask me edge cases on your language of choice in this. Don't come with a checkbox of things you want to hear. Have a discussion with me and actually pay attention to the things I have to say that aren't strictly part of the answer to your question.
I make what Glassdoor tells me is an average salary for my area and seniority, but I believe I could do better if I could muster up the courage to interview heavily and change jobs.
Whole process from scheduling the first interview to the job offer took 4 days. No white boarding, no obscure technical jargon. They met me (virtually), learned if I could code to their standards and made a decision. In my mind it was great. Via the same process they interviewed someone else I recommend and very professionally and politely turned them down, someone who in my mind was a pretty solid engineer but he just lacked the specific skill set they wanted. While creating the requested application, he knew he was not going to get the offer, which is great as you as a developer are then able to determine for yourself if you are a good fit for the role. No mess no fuss.
A few companies gave a simple api project or a timed hacker rank and those feel reasonable. An entire web app with database, rest api, front end and CI pipeline following production code best practices does not.
I’ve always gone through local recruiters that have relationships with the hiring manager. They know the salary range, the interview process, and where other candidates fail because they get feedback from both the candidate and the hiring manager.
You only need 1 offer to negotiate.
If I had already worked in the js stack they asked me to use the entire project build would have been 5 or 6 hours and that's only because I really wanted to impress, I could have knocked it out once my kids went to sleep.
But as in everything it depends on the person and their life circumstances, we all have different things going on.
As someone doing the hiring: the process should help order applicants by the value they will provide to the company, accounting for future growth and current abilities.
In both cases it should do so as quickly as possible, to avoid wasting my time.
Sadly, neither of these are really realistic, but starting from these first principles, you can see a few simple things that can improve the process: Early rejection in both directions, settings expectations, and that the interview is a two-way street and while some interview strategies may work in one direction, they will alienate the other side so quickly they are ineffective on the whole.
I have some ideas regarding the rest, but nothing that hasn't been mentioned elsewhere.
- Version control. Given a terminal (or Explorer with TortoiseSomething) and an existing project, make some simple changes and commit.
- Testing. Given a simple piece of code, its tests and either a bug report or feature request, explain at least in high level terms how to move forward.
- Code review. Give constructive feedback on how to improve any aspect of a diff.
These tasks indicate an understanding of code and facility with communication beyond the very basics, and you'll never finish learning them, so I believe they are appropriate for any non-entry level position. They are also sufficiently open ended that the candidate has to prioritize getting at least something done on all of them.
After that, unless they are completely hopeless I'd arrange coffee or lunch (on company money of course) with at least part of the team they'd be working with, so they can tell if they are at all compatible.
Needless to say, give the candidate plenty of options for a suitable time, let them know exactly how long it'll take, coordinate the time with the team, show up on time with a computer—which the admin team has wiped and you've just had to copy a few files onto—and find somewhere quiet and comfortable for them to work.
Does this really tell you anything about a candidate? If you sat me down in front of a terminal and said "here's a git shell, make a change and commit it" I'd fail the test immediately. I've used perforce for the last 9 years, and the terms are not the same, and the commands are _not_ discoverable. So you're eliminating anyone who doesn't know git basics (which can be taught in 5 minutes, or with a GUI of their choice), or you're giving a test for some criteria that you won't judge people on, so it's a waste of time.
> - Code review. Give constructive feedback on how to improve any aspect of a diff.
I like this idea actually. Have the person I'm interviewing review a piece of code (maybe a diff? I'm not sure about different diff formats, maybe it just needs to be some text files, or maybe they just have full internet access and the ability to download their own tools), and judge based on that.
I mean even if you don't know the command (or they are using a different version control system then you are used to) you can always check the help or man page. It's still a trivial task.
Disagree - You're testing can someone use the basics of git in a terminal.
> you can always check the help or man page.
Assuming you know how to do that, and what you're looking at. Say I'm sat down in front of a terminal, and I'm told "here's a terminal, make a change and commit it"
I type in "git", and I get: "usage: git" - If you're used to using shell tools, then sure you can make sense of it, If you're not, then you're done.
You said "commit" and the description for that in the "git" command is:
> grow, mark and tweak your common history
> commit Record changes to the repository
Right, I don't know what that means, but lets try this `git commit`
> fatal: not a git repository (or any of the parent directories): .git
Ok, No idea here. I somehow figure out that git init will give me a .git folder. Now that I've got that out of the way, I try git commit , I get:
> nothing added to commit but untracked files present
Ok, how do I track files? "git help" doesn't mention tracking files. I'll try "git commit hello.txt", which gives me:
> error: pathspec 'Hello.txt' did not match any file(s) known to git
I give up at this point really.
(by the way, I got this far by doing this walkthrough this morning, and googling "how to use git" - which told me the answer in 3 seconds).
Not knowhing how to use a terminal vcs, or knowing the commands to perform even the basics doesn't mean you can't use version control, it just means you can't use a terminal for git. Is my 9 years of C++ and perforce experience because I didn't know that commit was analogous to submit, or because I've used a graphical interface for all that time?
That's part of the test, to see if they're familiar with the command line and if they know how to open the man page. It's trying to weed out the people that can only work in the confines of an IDE and gui tools. That said I wouldn't expect anyone to know git from the man page, I would however expect anyone for a senior role to be familiar with what is a de-facto industry standard.
And source control in general is a great topic for interviews on both sides. Many devs (and companies) don't know what a branch is or what you'd use one for. Many companies make it hard/impossible to create feature branches, either by policy or crazy mono-repo stuff. Even their choice of SCM says a great deal about them, I'd avoid anyone that uses clear case or TFS.
That doesn't tell you anything about how good a programmer they are. I don't need to use a command line or man pages for 99.999% of my work, so I'm not going to waste time learning to use more tools.
> It's trying to weed out the people that can only work in the confines of an IDE and gui tools
Ah, so anyone who uses a terminal is superior to someone who uses an IDE or a GUI?
> I would however expect anyone for a senior role to be familiar with what is a de-facto industry standard.
In _your_ industry. As I've mentioned before, I use Perforce (which is standard in my industry).
> And source control in general is a great topic for interviews on both sides
Agreed, but asking somone to rattle off `git init git add . git commit -m"I can remember three lines"` doesn't tell you anything about how much they know about source control. Talk to them about branching/workflows to find out how much they know about source control, or let them use the tools they're comfortable with, but plonking someone in front of a terminal to rattle off some commands is the equivalent of looking for a "culture fit"
I don't want to offend but you sound like exactly the type of developer I try to filter out. IME there is a large correlation between at least rudimentary command line proficiency and being a good developer. Aside from that I want someone to know about the world outside of the IDE and what options are available, because a unix like environment offers far more power than an IDE.
I expect developers should be able to automate common tasks, work with build tools, grep through logs, remote into servers, debug on servers without an IDE installed and a million other things that are very common.
Maybe this doesn't apply to your specific industry but it does to everyone I've worked in to varying degrees. Not memorizing git I could forgive if you could explain branching workflows, but living life in the IDE I wouldn't.
Yet they carried mid/senior-level titles.
You should absolutely test for the things that you think are annoyingly trivial if this person is to be a close peer or a direct report, because your level of disappointment will be so much greater after that person becomes an employee.
I define "annoyingly trivial" as the things that you feel everyone should know "at this stage", and you would be annoyed at having a conversation about said topic for more than 5 minutes.
My personal opinion is Git definitely falls into that category.
I can count on one hand the number of times I've relied on man pages for descriptions. Using git as the example, compare https://man.cx/git-commit(1) to https://www.atlassian.com/git/tutorials/saving-changes for someone who has never used version control. One of them talks about saving, with examples. The other talks about storing an index in a log.
> My personal opinion is Git definitely falls into that category.
There are plenty of developers with working knowledges of branching/merging workflows, version control, using visual tools, or non-git tools.
Or even better, this one about git: https://xkcd.com/1597/
It's good to use version control, but I wouldn't rate it too high in judging developer competence.
And for the record, I try to interview people on their strengths, giving them the opportunity to show off what they know about the area where their greatest interest or experience lies. A shocking percentage of people (>50%) nevertheless fail to demonstrate even basic knowledge of syntax or language features in the language of their own choice. This is without even considering actual workplace skills like git, PR review, tickets, etc. as discussed above.
1) Casual talk. Does this person have basic communication skills and are they someone you want to deal with regularly for a long time?
2) Give them an actual coding project to do. Pay them for it. Pretend it's just like real work (it should be).
3) Review their code with them. (both to make sure they are the ones who wrote it and to explore their thinking process)
I’ve been doing my tech stack long enough that I can smell bullshit and for me I need to know that you can be a good communicator (tech stack is teachable).
Problem solving is a fairly universal thought experience not limited to writing code. The goal is to ascertain whether the candidate can break down the complexity of a problem into simple steps, organize their thoughts into a clear flow, communicate clearly, and finally recommend a valid solution.
You don't need to test whether the candidate can actually write code, because this is built into the nature of the exercise as qualified by the feasibility of the proposed solution. This exercise also implicitly tests confidence, creativity, experience, and approach style.
Most importantly though, it separates the competent from the incompetent. No amount of framework foolishness and dependency baggage will communicate a solution for you.
I recommend trying to make the cost/benefit more symmetric. One way of doing this, is offering compensation for real work. Essentially, hire interviewees to do a small amount of useful (a day?) and evaluate them based on that.
This isn’t going to work for all interviewees, but where possible it feels like a better way of doing things to me.
It's not really worth it to interview nowadays. Practicing leetcode for months, flying, doing all day onsites, when all the company has to do is interview you for a few hours.
We've been issuing take-homes lately, and we ask that candidates spend no more than 3 hours on them. Well, it turns out almost no one spends only 4 hours on them... The most recent candidate we interviewed spent about 3 days on theirs judging from the commit logs. It was a fully fleshed out application with unit tests and everything. After all that they still have to pass the all day on-site, lol. Anyway I spend all of 2 minutes looking at each submission
I really wish there were a better signal.
I was thinking the opposite - that no employed engineer is going to bother with a multi-day paid contract for a new position, so the only engineers you will attract are those without a job and no pending interviews.
Remember Joel Splosky’s classic observation: all the good programmers already have good jobs. The ones on the job market who apply to jobs are usually the bad ones.
I am more than happy to take the time off just for the opportunity. If it doesn't go well or the position is not a good fit then I go back to my current job.
Sure, they don’t know the codebase, but they won’t know the codebase that they’ll be working on in a month if hired either. I want to see how they think about creating software and whether they notice potential defects or tricky areas.
My apologies in advance that this may not be the response you were looking forward to on forum question.
But I discovered an older post online on the Hacker News site some time ago for a Remote Junior Programmer/Assistant at Luma. I noticed you were looking to make your first hire and find someone that has some knowledge in technologies such as Python and Linux.
Well, I had completely forgot about viewing it and so I was checking out Hacker News job board today and saw it was for a remote Part/Full-time opening available at the Luma in the NYC area.
Well I'm someone who enjoys coding and learning with Python as far as learning purposes/hobbies go. Plus, I'm someone who has been going the self taught route trying to break into the IT field/Python Development world. I also have interests in some of the things mentioned in the post such as finance/trading and business.
But, more than anything.. I'd looking to further learn and grow in my skills as far as Python development goes. I don't have much experience with Django Web development but feel I can learn and pick up on it as well as with any other technology requirements quickly.
After quickly reading your post on Hacker News I was encouraged and interested to reach out and contact you for more info.
So with that said, here I am..and I wanted to inquire to find out if this opening is available? Or has it been filled? Also would you allow for training of junior developers to get up to speed? I had to ask because but I wasn't sure if you meant junior or for more seasoned developers.
Also, you have a contact number to learn more about this position or information on the things you require in regards to the nature of the job to increase one's chances to be a part of your startup team? Any help in this matter will be greatly appreciated. Thank you
Feel free to contact me at: firstname.lastname@example.org
Of course in order to get there would require a working certification system. Requiring candidates to write fizzbuzz in person over and over is a waste of your time and theirs, but it exists because nobody's managed to crack a "fizzbuzz certified" system that employers actually trust for this. The fraud pressure is quite high and nobody wants to pay for it if there is a possibility that someone might benefit - 1000 candidates applying to 100 employers results in a huge amount of waste, but requiring the candidates to pay for certification cuts out a lot of good candidates, and paying on their behalf and allowing someone else to use that result is seen as unacceptable.
I don't ask FizzBuzz because many expect it, but I ask equivalently simple questions. I have the following criteria for an interview code question
1. Have an obvious KISS implementation.
2. Ability to modify requirement so the KISS no longer works.
I expect 2. to enable us to gauge their thinking. I don't expect a working solution to it, but they should be able to identify why it does not work anymore and outline ideas to solve it. This part is a conversation where we can lead them if they are not good at controlling the conversation.
Some examples are things like:
- Walking the candidate through an outdated API (that the candidate isn't familiar with, but should be able to understand given the nature of the job)
- Walk the candidate through code that converts a database query into objects without an ORM. (Candidates who can't do this are incompetent. Really.)
- Discuss commonly-known details of exceptions / error handling in the language that the job is for
- Discuss commonly-known details of memory management in the language that the job is for
- Discuss API choice tradeoffs in an API that the candidate should be familiar with. (I like to pick serialization APIs built into the library in the language we will use.)
- I try to emphasize how I would do with my interview at the candidate's level of experience
- I have 2nd chances, and will usually stay on the phone for the scheduled interview out of respect. (There are still certain points where I will cut an interview short.)
- Make a decision to hire quickly.
Usually works well.
Signs to reject someone:
Figuring out how to use the teleconference software is an unofficial part of my interview. (Most of my interviews are conducted via teleconference because I'm on a global team.) There are candidates who take 15 minutes to figure out that they can type their phone number into the teleconference web site and it will call them back.
Candidates who want to answer a different question, or keep asking, "why can't I use XYZ" technique usually aren't hired. Again, the purpose is to have a discussion about code and demonstrate understanding of the discussion. I don't want to hire someone who can't adapt when the correct solution to the problem is some tool / design pattern / API / ect that isn't the candidates first choice.
If the role does not fit the candidate, certainly it's advantageous to both that the candidate does not get hired.
that's ideal, but in practice i think employers and candidates are conditioned (perhaps by prior less professional employment? or the market as a whole?) to think that it is the candidate's job to trick their way into the company, and the employer's job to guard against the unqualified sneaking in.
And sadly I haven't once seen it in action. I really wish I could go through such process, or implementing it in whatever company I work at (or create).
It's actually kind of interesting doing this from the hiring side because it became evident that there are a lot of relative "bargains" out there - excellent developers who do poorly in whiteboarding and chatty interviews so have lowered salary expectations but who are nonetheless very good at their actual job.
On the other hand, I dislike interviews where interviewers come up with a question-answer sequence. So, if you misspoke even a single word they outright reject you. And mostly, I find this in places like TCS, Wipro, Infosys etc.
It's also anonymous until the 3rd stage, so you (by and large) interview remotely and without anyone knowing your name / gender.
(this dates from 2015)
It’s actually building something in my element, IDE, docs, just like real life. You get a chance to show you care with the details.
A whiteboard interview is not like RL as your entire career is on the line so shouldn’t be the main KPI, your brain is not working as usual. I don’t even hate presenting to groups or anything, although I do hate presenting unprepared.
Start with a very, very simple initial phone screen or take-home test, intended to basically verify whether the candidate can write any code, at all. Max 1 hour, weeds out more people than you'd think.
For the first in-house interview, ask the candidate to code up a problem that is representative of your company's work and requires coding a significant amount, ideally 100+ LOC. The problem should not require any major leaps of intuition, dynamic programming, or recursion – all of these are areas where people do way worse when they're nervous, and this is an engineering interview not special forces training. Let them bring their own laptop, give them the prompt, and have them code, although they can ask the interviewer questions at any time. When they're done, go over the question in detail with the expectation that their code compiles and runs, discuss extensions, etc. Max 1 hour. This interview should answer the binary question "can this person promptly produce meaningful working code and discuss it intelligently?"
For the next in-house interview, do a deep dive into a technical project that the candidate worked on that they're proud of. They describe it and you ask questions. Keep asking questions, especially getting at the "why" behind different decisions, for as long as you can – you're trying to get to the borders of their knowledge and intelligence. Look for mastery of the area, thoughtful decisions, and communication skills. Max 1 hour. This interview should answer the question "is this person a thoughtful, effective, smart contributor on a project?" A good answer should make you think "damn, that's really smart, I wonder if I would have thought of that?" at least once.
End with a final behavioral interview, intended to sell the candidate. This is also a last gut check on whether they're insane, dangerous to themselves or others, extremely arrogant, etc. Also use this time to ask the candidate questions about what really matters to them to improve your closing rate. 30 minutes, and can be combined with the step above.
I've liked this system, YMMV. It's a relatively efficient process, doesn't have weird tricks, and based upon a longterm analysis of candidate outcomes was quite effective (this included an analysis of candidates whom we rejected and who rejected us).
1. Test language-specific knowledge - it's not a bad thing to not know all the small quirks or details of a language since they're generally not too useful, but when someone does know them it tends to be a good sign that they really enjoy coding and learning. And of course there is a certain minimum amount of knowledge that should be required - listing Java as "proficient" on your resume while not knowing the difference between abstract classes and interfaces might indicate a problem.
2. Design question - just a simple toy problem like "how would you design the book management system for a library?". It's easy to use such a problem to probe into some different aspects of programming like concurrency and databases, just to see how much the person knows.
3. A not-crazy-hard algorithms questions - people often say algo questions are irrelevant to actual work, and I agree with that. But I think algorithms are really a core part of the CS curriculum at every school, so getting completely stuck on a medium-difficulty algorithm question should raise some flags.
4. Resume-specific things - it's always nice for an interviewer to show that they've actually read your resume, and it can be a good way to convey some strengths that aren't otherwise evident.
I guess my philosophy is to interview in a way that can test the depth of a candidate's knowledge while not being obnoxiously tedious or memorization-focused. i.e. someone who has written a lot of production code should do better in an interview than someone who's just memorized every problem in Cracking the Coding Interview. So, ideally with the screening round clearing the first part (language-specific knowledge), and then 2 or 3 subsequent interviews that cover design and algos.
Most people who learn algos & DS never have to implement them on the job, and often haven’t actually written their own since their CS classes. If the question is any deeper than “what are some good and bad choices of algorithms for this problem”, you’re selecting more for who has just graduated or has been grinding leetcode and doesn’t have performance anxiety than you are for competence, expertise, and productivity.
I know folks who got rejected from the leetcode companies and went on to build major products from scratch. The 0.5% of work that requires them to have to implement something like this manually, they’ll just grab their CLRS book and refresh their memory long enough to solve the problem and probably forget it again a week later.
I also know folks at these companies who started fresh out of college, passed that style of interview, worked there for 7-10 years, moved up to senior or higher in the engineering track, and openly admit that they would be a deer in the headlights if they got any of the interview questions. Without going back to study and practice those, they know wouldn’t pass an interview for a junior dev, let alone one for their current job (where they are currently considered high performers).
The big problem is that this style of interview filters for people who pass this style of interview, and then those new hires are asked to interview others, so they interviewed in the style that they were given and passed.
The kind of question I had in mind would be along the lines of "say you have a tree, how could you find the nearest common ancestor of two nodes in that tree?". I actually got this question during an interview recently, and I felt it was of a suitable difficulty where I could figure out a solution without any specific knowledge other than what a tree is (which most developers should know).
I'm sure I'm still young and naive, but my line of thinking is that a question like that is good for having a person show their thought process when figuring out a problem they haven't seen before. Design questions are a bit different because they tend to just require a rehash of a codebase you've worked with before.
Algorithms, who cares. Maybe as a tie breaker.
I think you can quickly get a sense of depth, potential, curiosity, and passion if you can get them talking about quirks and their opinions on those quirks.
For example, if someone says that they once attached a debugger to the JVM to confirm a pathological issue arising from benign looking Java code and found that the JIT compiler falls apart when a certain construct is used, I’m probably going to hire that person, even if we have no Java code in the company - assuming I can also confirm that they are productive and don’t just waste time going down deep rabbit holes.
In my experience curious developers who like to dive deep light up and love discussing that kind of stuff.
Testing random trivia is dumb.
Coworker and manager feedback, and my compensation over the years, say I'm pretty damn good at adding value despite this. I've been considered the "smart one" and the go-to guy for hard or weird problems. But I will struggle to produce something that will even come close to compiling on your whiteboard, in a language I wrote 300 useful lines of yesterday, unless I spend a couple days basically flash-card drilling for your test. Go figure.
No one is ever expected to solve an unfamiliar problem in a 60 minute meeting. Generally a story/feature with technological unknowns are pointed for several days and are in the developers current wheel house, yet we expect a candidate to use most of their energy and facilities in a social situation with people they’re unfamiliar with and solve whatever random algorithm, riddle or puzzle that you Googled the answer to before the interview.
Ask me about my past projects and decisions I made and mistakes I made and what I’d do differently. Sit down with me and actually code with me and get a feel for what that’s like. Ask me to how I’d design some actually realistic system, and drill into the details of each.
I shouldn’t have to train to pass an interview. Prepare, yes. Practice a bunch of problems unrelated to the job at hand, no thank you.
2) > 45 min < 3 hours long pairing test or test otherwise done in the presence of the interviewer.
3) Minimal setting up development environments, frameworks or test environments. No boilerplate should be required during the test.
4) The test is improved iteratively. Catch bugs during each interview process and fix them in subsequent interviews.
(the following is just inspired by your bullet)
The job of a developer consists of problem solving something they have never encountered before (under some pressure), communicating to other developers, and demonstrating knowledge in various combinations and amounts over time. Startup developers are the hardest spots to fill and most delicate, so I tend to always think as if I'm hiring for a startup.
Talking through a few whiteboard problems and talking about tools they have used (why and how) are sufficient. 30-45 minutes. I don't subscribe to adding more elements (environments, tools, test frameworks, etc). 3 people, 1 from the destination team, 1 from another team (if possible) and the immediate manager.
Startups to multinational corporations, I'm still convinced this is the best way over the last 20 years.
Interviews with higher management is just a redundant filter and another reason for the management to justify their existence. It's a job smell that every company has and is obviously prejudicial in every regard (trying to tease personal details or redundant workflow experience and opinions). It's a bad practice.
I've done interviews with a whiteboarding "tell me about stuff you've worked on" component followed by a test.
It thought it gave a weak signal about the skills of the candidate. I canned it eventually because the test gave a very strong signal and it wasn't telling me anything the test didn't.
>Interviews with higher management is just a redundant filter and another reason for the management to justify their existence.
I agree. I had suspicions in a previous company that this upper management "team fit" interview was adding a race/nationality filter that ended up with good candidates getting dropped.
I'm really frustrated when these kind of problems are presented as "whiteboard" problems. It's not constructive and discouraging to potential employees. When something as simple as "implement X without using Y" or "fix this program with pseudocode" are practically useful experiments and more indicative of what developers are doing to be doing ALONE every day. If there's inefficiency, code reviews and planning and integration all reveal that and you can pull in specialized resources if you really need it.
Part of the problem I had with this was that I didn't feel like it was even a dev challenge. I've hand coded reports before and that has always led to a world of pain. I also felt like it was a data sciences challenge, not a dev challenge, and my data science is really rusty.
I spent most of the 1-4 hours they said most people complete it in, just thinking about the problem. "If I had to solve this problem in my company, first thing I'd do is look at Crystal Reports. The last thing I'd do is open a file and type "import sqlalchemy"."
I set up a repo that did all the operations stuff (remember, it was a Ops job I was applying for), and put together deployment parts to set up a test system and load the data into the database, configure everything, etc...
A few weeks later I finally got the parts all to work and was able to solve the problem in an 80x25 screen worth of SQL. I suspect I was the only applicant that solved it in SQL.
1. "Imagine you are building a simple Gmail clone (search header, sidebar, action navbar, main content area). Walk me through how you would approach and implement this."
(20 minutes: What architecture choices do I make? What tradeoffs am I comfortable making with this? Do I ask questions about the audience? How do I handle state management? How do I approach tooling for this? How do I approach testing for this? Lots of specific questions along the way..)
2. "Tell me about an interesting article you read recently on a tech topic? Which resources do you use to stay current in the front-end space?"
(10 minutes; Am I committed to learning and staying abreast on an ever-changing landscape. Can I impress me with quality resources I'm in tune to? Also, can I effectively convey ideas at a high level; can I critique it and walk around the topic from various vantage points.)
3. "Could you pull up your github and walk me through your last few public commits."
(15 minutes: Gives me a chance to see my code, talk about the process of writing it? Are there tests? etc)
4. "Would you mind code reviewing this [FizzBuzz-like] code and tests? Then, what would your next iteration on it be?"
(15 minutes; Am I a good team player? Can I communicate effectively? Can I spot areas for improvement?)
5. "Finally, could you provide me with a list of past/current co-workers"
(0 minutes; With this I will be able to assess what my peers thought of me? Work ethic? Pleasure to work with? Ego? Best qualities? Shortcoming?)
I suppose if the first question is switched, this could be used for any position.
The candidate selects 3 questions to answer from the set of questions. We expect them to take ~15 minutes to answer each question using the English language.
Obviously not a complete answer, but I think it's a component that isn't being mentioned here. Disclaiming that I don't have much experience (still in university).
One of my internships went from "applying -> interviewing -> accepting" in under 24 hours. Impressed the hell out of me. It was really the reason why I ended up accepting their job. 24 hours is obviously abnormally fast, but a week or two doesn't need to be.
One of my friends ended up rejecting an offer both from Google and Apple because the interview process took too long with no responses. They got a good offer in the mean time, and after waiting over a month for a response they decided they had to go with it. (They followed through with the rest of the Google/Apple interview process anyways for the experience... which was basically just host matching and getting an offer).