ML system design

May 18 5 Comments

Any good comprehensive resources targeted towards ML design interview rounds? Looking for something similar to Grokking/donnamartin like refs.

There are so many common applications that are being asked like designing recommendation systems (products, dates, jobs), ranking/relevance models(news feed, search), matching problems(matching dates, driver-rider, gamer match) etc but am surprised by lack of resources online. Please help if you know any. Thanks

comments

Want to comment? LOG IN or SIGN UP
TOP 5 Comments
  • Sizmek athV68
    If it’s an ML role, then you need to know traditional ML algorithm and when to use them and how to build an infrastructure for training / evaluation / tracking analytics etc. If it’s for an ML infrastructure role, then you should expect regular system design interviews.
    May 19 1
    • OP
      Yes. Any resources online where multiple approaches are presented and trade offs are discussed? Company blogs seems to be a good place to start but they may be too specific given their problem.

      Ideally looking for an overview of different design components involved, different ways to think about designing them, pros/cons etc
      May 19
  • Accenture / IT faoehfpq
    one of the comprehensive materials out there https://github.com/rjurney/Agile_Data_Code_2 , and just search on youtube "machine learning system design"
    May 19 0
  • Adobe
    romeleet

    Adobe

    PRE
    Intuit, Autodesk
    BIO
    Keep a left Sucker
    romeleetmore
    Blind is cloud heavy OP, unfortunately you wont get good advice for this. What is your specialization in ML?
    May 19 0
  • Sizmek athV68
    Honestly I guess grokking would be good for that. A system doesn’t drastically change just because it is ML based. What’s happening at MZ that makes you consider a change.
    May 19 0

Salary
Comparison

    Real time salary information from verified employees