4 ms·

Ask HN: Really hard take home interview, what to do?

I got offered an interview at the research division of a tech company. They sent a take home test that am supposed to work on. It involves building a backend with some sort of machine learning/AI, basically pretty advanced stuff. I only have a weekend to do it, since I have a job and can't work on it on a week day. I would love to work in a research division but don't think I can deliver sth meaningful within such a short time span. I've also been brushing up my algo skills in readiness for a gruelling on-site interview so that's another time sink. I am very comfortable with interviews but I don't feel confident about this one. What should I do?

koolba 10y ago
You can either be honest, try your best, and submit whatever you're able to produce. At that point it'd be up to them to decide if it makes sense for you to join their team. Or, you can be like everyone else and enlist help from others to complete it. On paper you'll look better (initially) as you'll have passed whatever test they've got but it could balloon into problems down the road as you're exposed. Essentially you'd be in a "fake it till you make it" situation. I suggest being honest.
- concurrent_eng 10y ago
  I barely have any machine learning experience, I would need hours of research to come up with sth good enough to show.
  - davelnewton 10y ago
    Weekends have hours. The internet and books have enough to get going. So far it sounds like you already have your answer.
  - smt88 10y ago
    Is it possible you're not qualified for this job, and you'd be more successful (and happier) at a job with better fit?
    - concurrent_eng 10y ago
      From my own estimation, and my current boss, I'd say am a competent Software Engineer, so with time I would be very productive in this job. I've also been curious about machine learning and this could be a potentially good opportunity.
      - smt88 10y ago
        That makes sense if the employer is OK with you learning ML at the company. It doesn't make much sense if they want to hire someone already experienced in ML.
davelnewton 10y ago
I'm not sure I understand the question. What do you think the options are? I only see a few: * Give it your best shot * Bail and tell them you can't/won't do it * Tell them you need more time to show you what you're capable of Not knowing what the test actually involves it's difficult to provide any meaningful advice. There are a lot of packaged solutions for ML these days, and many of them are relatively easy to get started with. There are even books that walk you through the process. If I were doing it I'd just resign myself to the fact that a weekend, even assuming you could be productive for 48h straight, is only 48h. If I really wanted the job I'd take a good portion of those to giving it my best shot, but I'd sleep and take breaks to avoid the burnout and poor quality that would come through. If, on Monday morning, I felt comfortable enough with my result, I'd send it to them and say "this is what I had time to do". If I didn't feel like it was at the caliber I believed they'd accept, I'd tell them I took my best shot, but it wasn't up to my own standards, and ask what they thought the next steps should be. I'd be honest about wanting the job, I'd be honest about the holes in my knowledge, and I'd ask for advice.
- concurrent_eng 10y ago
  My options are: * Give it my best shot (but is it worth it? I could as well spend so much time on it and still get declined). * Bail on it (am not desperate for the job). The task is up to me to settle on sth, what's being looked for is extraction of meaningful insight from things like images, audio etc, and an illustration of the extracted info.

koolba 10y ago

You can either be honest, try your best, and submit whatever you're able to produce. At that point it'd be up to them to decide if it makes sense for you to join their team. Or, you can be like everyone else and enlist help from others to complete it. On paper you'll look better (initially) as you'll have passed whatever test they've got but it could balloon into problems down the road as you're exposed. Essentially you'd be in a "fake it till you make it" situation. I suggest being honest.

concurrent_eng 10y ago
I barely have any machine learning experience, I would need hours of research to come up with sth good enough to show.
- davelnewton 10y ago
  Weekends have hours. The internet and books have enough to get going. So far it sounds like you already have your answer.
- smt88 10y ago
  Is it possible you're not qualified for this job, and you'd be more successful (and happier) at a job with better fit?
  - concurrent_eng 10y ago
    From my own estimation, and my current boss, I'd say am a competent Software Engineer, so with time I would be very productive in this job. I've also been curious about machine learning and this could be a potentially good opportunity.
    - smt88 10y ago
      That makes sense if the employer is OK with you learning ML at the company. It doesn't make much sense if they want to hire someone already experienced in ML.

davelnewton 10y ago

I'm not sure I understand the question. What do you think the options are? I only see a few: * Give it your best shot * Bail and tell them you can't/won't do it * Tell them you need more time to show you what you're capable of Not knowing what the test actually involves it's difficult to provide any meaningful advice. There are a lot of packaged solutions for ML these days, and many of them are relatively easy to get started with. There are even books that walk you through the process. If I were doing it I'd just resign myself to the fact that a weekend, even assuming you could be productive for 48h straight, is only 48h. If I really wanted the job I'd take a good portion of those to giving it my best shot, but I'd sleep and take breaks to avoid the burnout and poor quality that would come through. If, on Monday morning, I felt comfortable enough with my result, I'd send it to them and say "this is what I had time to do". If I didn't feel like it was at the caliber I believed they'd accept, I'd tell them I took my best shot, but it wasn't up to my own standards, and ask what they thought the next steps should be. I'd be honest about wanting the job, I'd be honest about the holes in my knowledge, and I'd ask for advice.

concurrent_eng 10y ago
My options are: * Give it my best shot (but is it worth it? I could as well spend so much time on it and still get declined). * Bail on it (am not desperate for the job). The task is up to me to settle on sth, what's being looked for is extraction of meaningful insight from things like images, audio etc, and an illustration of the extracted info.