LinkedIn Data Science Team

Chegg snt124
May 23 18 Comments

Are data scientists at LinkedIn more ML DS roles or analytics, or a mixture of both? Wondering if the roles really require strong skills in python/ML

comments

Want to comment? LOG IN or SIGN UP
TOP 18 Comments
  • LinkedIn MxAu52
    Ds at LI does not develop ML models- there is a track within DS which is more ML focused but for the most part it’s more analytics. If you’re looking for core ML work then interview for the ML/AI track
    May 24 1
    • Intuit tgfi
      Why do they ask all those stats questions then?
      Jun 11
  • Amazon / Data pozbuw
    I interviewed for a ds position at sunnyvale, and yes it was analytics, ie not writing production codes for ml or ai products, but more focused on supporting decision making with data. The interview was not easy btw, there were product case study questions, advanced stats topics like whats your favorite kernal function, describe a difference between l1 and l2 regularization and talk about when you used either and why, but no r or python coding, but leetcode medium sql questions.
    May 23 9
    • Amazon / Data pozbuw
      Wow you improvised on penalization on the spot? Thats pretty impressive, you must have super solid math intuitions
      May 23
    • Amazon / Data pozbuw
      OP, it was not a sr ds i interviewed for, it was a level below. The thing is I knew what L1 and L2 regularizations were both conceptually and had a good amount of experiences implementing them in R, but they asked me to write out the formula or show it graphically, which at the time I didnt have a full grasp on, and they made a big deal out of this in the feedback.
      May 23
    • Goldman Sachs fake engineer
      I drew the graphs just fine with the equal loss contours and Uber still didn't give me an offer...Lol. Maybe they were disappointed I didn't spew it from memory, or that I said in practice you would try both on the data since the end result is kind of similar.
      May 23
    • Amazon / Data pozbuw
      They are stupid not to appreciate that you were smart enough to come up with it on the spot. Just curious, what do you do at GS?
      May 23
    • Goldman Sachs fake engineer
      I babysit mounds of technical debt...:( It would be fine if the comp were above average, but it's the opposite, so trying to gtfo

      Anyways back on topic, it's true a lot of product analytics roles won't ask this stuff. In those cases, you should be good at the product case question BS. I think it's mainly about following the consulting interview format where you give your "structured" response, since that's what makes people feel good.

      There really isn't an excuse for not memorizing the answers to a bunch of the really standard stats questions though. All the stuff you can find in a textbook about linear regression (Gauss Markov, regularization, standard errors, t stats), all those probability/combo brainteasers, those "what is the variance of blah conditioned on blah" questions, etc.
      May 23
  • Goldman Sachs fake engineer
    When I applied there were two different job titles, machine learning engineer and data scientist. I'm pretty sure data science is much more analytics focused. I was only asked soft questions during the phone interview for data scientist and didn't pass.
    May 23 5
    • Chegg snt124
      OP
      Ty. Any idea why you didn’t pass? Do you think you messed up the soft questions or were you surprised that you didn’t pass?
      May 23
    • Goldman Sachs fake engineer
      Not surprised I didn't pass, I was asked some weird product question about LinkedIn chat or something and I really had trouble bullshitting. I wish I could have said "I think these are terrible ideas we should be thinking about something else".

      I quickly realized I suck at product questions if it's not about a product/industry I know well (even for that, I didn't pass a phone screen with Quora despite being a power user and thinking I crushed it) , and decided I should try interviewing for engineering positions (elsewhere) instead.
      May 23
    • LinkedIn Thsiswd
      Fwiw if you gave a opinion with a quantitative backing you would have done well. As much as these interviews are about bullshitting , they are really about logically thought processes and whether one can approach problems with a quantitative mindset
      May 23
    • Goldman Sachs fake engineer
      Perhaps so. At LinkedIn I really had no idea what to say because the question seemed so weird to me. The Quora interview I gave them a lot of explicit metrics and optimizations and they still weren't happy, so idk.
      May 23
    • LinkedIn flurple
      “I think these are terrible ideas we should be thinking about something else" could be a pretty good answer to an interview q, if you have some better ideas for what to do instead.
      May 24