4 ms·
I think the real issue is the misapplication of take-homes. If after 3-4 rounds of systems interviewing, the company came back and said: "Hey we like you, and
by ilc 3y ago
I think the real issue is the misapplication of take-homes.
If after 3-4 rounds of systems interviewing, the company came back and said: "Hey we like you, and we're interested in making an offer, but we want to check your skills..." Yeah, I'd do the assessment. Especially if I did 1 round of 45m coding. Without that, the company can't verify that I have "FizzBuzz" level skills, and that I didn't cheat on the assessment.
When it is upfront, it is rarely worth it on either side. Companies lose a large fraction of the engineers applying into their funnel, and honestly as an engineer, it is rarely worth my time. I can spend that time interviewing at other companies that respect my time.
As with many things in life: It is when, and how things are done. Not what things are done that often matters.
- arp242 3y agoYeah, this is also my issue. You get a "take-home" after a short 15 minute intro call where some HR person basically just repeat what's on their homepage and job listing and where you repeat what's on your CV (I have a theory that HR people can't read, all observed evidence points in that direction anyway) and then you're given an 8 hour take-home. Sometimes you don't even get the intro call and just an email. Even just a "hey, we're genuinely interested and you made it to the shortlist" or something to that effect would just make a world of difference. Now I have no idea if I'm on any sort of shortlist or if they're just shotgunning this to everyone. I tried asking a few times, but I always got such vagueries in return that I might as well not have asked, so I stopped bothering. Would also help if people made the tests at least a little bit interesting. Now it's either "do this boring CRUD thing you've done 1000 times already" or "solve several NP-hard problems that have been subject of Turing awards within an hour". The last time I spent quite a bit of time on one of these things because it was interesting, and even put the result on GitHub (with no mention that it was for a take-home test, it "just" looks like any other project, also slightly modified to fit my needs). Then again, what's "interesting" for one person is "boring" for another.
- ilc 3y agoI've handed fully containerized code, with 100% code coverage. (Separate entry point from default to run the tests.) Then been told no. I suspect the people weren't used to having someone hand in a basically 100% guaranteed to work program ;). (I do know 100% code coverage doesn't mean working. But the docker container helps a good bit in making sure the code will run in the environment I wrote it for.)
- arp242 3y agoEvery time I've reviewed these sort of code tests I've never even run the code. I just looked if the approach looked alright, and to be honest I don't even care if it gives a slightly wrong answer because everyone can make an error; this is why we got code reviews and such. It's like physics and math classes at school; at least at my school an error in arithmetic was only a small minus if your general approach was okay (one reason I'm not in favour of computerized testing/ratings, because they can only look at the answer). I don't know what your specific code test was about so I can't really judge, but generally I've considered overcomplicated solutions to be a minus. "I asked you to write a simple CLI and I get some container environment" is not something I would consider positive anyway.
- ilc 3y agoYou think simple CLI. I think: I'm writing this in Python, most Python setups are a giant mess. By putting it in a container, I can guarantee it'll run right on their system. This is LESS complex. If my employer considers "docker run" when given full instructions complex. I guess I'm not a fit.