Google Onsite MLE L4 Advice

New / Eng
ershman

New Eng

ershman
Jun 23, 2021 16 Comments

My onsite for Google consists of 3 coding rounds, 1 ML domain knowledge and 1 behavioral. No system design.

I hear the bar for MLE’s coding rounds are a bit lower than traditional SWE, should my prep be focused more focused on Graphs, trees, dp? Or does anything still go including stuff like linked lists n bitwise.

Also any advice on what to review for ML domain knowledge interview? Just the basics on ML/DL?

YOE - 3+
TC - 150k+ in Chicago

comments

Want to comment? LOG IN or SIGN UP
TOP 16 Comments
  • Apple
    tf67gp

    Go to company page Apple

    tf67gp
    Did you get to pick a specific domain? It could be CV/NLP/general ML etc, the domain interview is related to it.

    Anything goes for coding, but I did feel like my coding rounds weren't great and I still got an offer for L5
    Jun 23, 2021 4
    • Apple
      tf67gp

      Go to company page Apple

      tf67gp
      Cool, fyi my domain round was also system design-ish (in computer vision), so I would recommend preparing a little on recommendation system + NLP, all the new large language models and how they fit in with problems like next world prediction keyboard, search etc. Know some ML design principles (like inverted indexing for fast retrieval). It can prevent down leveling as well.

      I got one hard (did not do finish solving it optimally), rest were medium. It was either one question and a tricky follow up, or two questions, so don't panic if you get asked just one question.
      Jun 23, 2021
    • New / Eng
      ershman

      New Eng

      ershman
      OP
      Awesome yeah ill definitely prepare for that, thanks for the detailed response!
      Jun 23, 2021
  • New
    n7gd97v

    New

    n7gd97v
    No difference in coding rounds - expect the same level as general SWE.

    It is up to HC to decide if your interview performance meets the bar. Since there is only one SWE ladder (doesn’t differentiate frontend/backend/mobile/ML), HC would expect similar performance too.
    Jun 23, 2021 1
  • Coding rounds will be standard. I was asked 2 hards and 3 mediums across 3 coding rounds. 2 on DP (mediums), 1 on hash map / interval sets (hard), 1 on Bin Search (medium) and 1 on monotonic queues (hard)
    The ML domain round consisted of rec sys question (just the basics, equations, training, metrics and such)

    For ML review, I went through the Deep learning book (if you have time and/or notes) and some rec sys notes
    Jun 23, 2021 5
  • This comment was deleted by original commenter.

    • It really depends. Some of the MLEs in my team (AdsML) do pretty cool prototyping/SOTA ideas implementation works.
      Ofc there are also the kind of roles you mentioned.
      Jul 2, 2021
  • Roche
    GmpB04

    Go to company page Roche

    GmpB04
    Hey what kind of work experience do you have? Do u have a PhD
    Jun 23, 2021 1
    • New / Eng
      ershman

      New Eng

      ershman
      OP
      I have an MS in Applied Math and work as a Sr MLE currently
      Jun 23, 2021