Data Scientist vs Machine Learning Engineer

BASF
FSAB

Go to company page BASF

FSAB
Aug 3, 2020 59 Comments

Have been thinking changing jobs recently. For future development, which one has more potential, DS or MLE?

542 PARTICIPANTS SELECT ONLY ONE ANSWER
VOTE VIEW RESULT

comments

Want to comment? LOG IN or SIGN UP
TOP 59 Comments
  • Are’s they basically the same?
    Aug 3, 2020 19
    • Yeah, pretty much all FANNG Senior Data Scientists are full stacked. Haven’t heard anybody straight up comparing Data Scientist as simply an analyst. I do know some companies that has no idea what DS is but simply want to hire an analyst put out Data Scientist position. It really is up to the job seeker to read the job description to decipher how that company interpret the role of data scientist. A lot of times DS and MLE is interchangeable.
      Aug 6, 2020
    • Prudential
      bigherc

      Go to company page Prudential

      PRE
      Morgan Stanley
      bigherc
      Yeah I saw Walmart’s DS openings and saw it was just a business analyst position. I was in BA but finally made it into DS and now I’m building models and putting it into production
      Sep 26, 2020
  • Facebook
    spysausage

    Go to company page Facebook

    spysausage
    DS are, well, scientists. They should be savvy with experimental design and causal analysis that drive large-scale strategy. A priestly caste.

    MLEs are the manual labor for implementing the DS’s insights at scale. Important, sure, but theoretically a little more proletarian
    Aug 4, 2020 13
    • Apple
      mTpL41

      Go to company page Apple

      mTpL41
      You really have a very narrow world view of what 'core ML' is and what MLEs work on. A/B testing and experimental design to bring out correlations doesn't make you a scientist, it makes you a statistician. A statistician who makes visualizations is a good way to sum up DS. I don't see DS publish papers at NeurIPS or ICLR, and no KDD doesn't count, lmao.

      As far as MLE, I'm not sure what your org considers MLE. But every MLE has to know all the topics you listed. I've worked at FAIR and also collaborated with AML folks before and we had some MLE who wrote papers and collaborate well with scientists.
      Aug 10, 2020
    • @syntaxbugs Yes, you're so smart, smarter and better than all DS. They don't know anything, only you are useful and worthy of respect. From your posts, it's clear you have no idea what you're talking about. Keep putting down other people so you can feel big and important little man.
      Aug 20, 2020
  • BASF
    FSAB

    Go to company page BASF

    FSAB
    OP
    From comments, it seems DS has more potential. But the votes show 70% chose MLE. Very interesting! I'm a DS now, not quite familiar with the whole model deployment pipeline. So I am planning to find a MLE position to gain some experiences. Then I think I will really be familiar with the end to end ML model development. How do you all think of my strategy? Thanks!
    Aug 8, 2020 5
    • Oracle
      lwnb8

      Go to company page Oracle

      lwnb8
      @SFDC Yes and no. NLP and CV are specialties in DL that go pretty deep, and a generic intro to DL isn't necessarily enough. There's a lot research that's specific to certain types and structure of data. Jobs exist for people who focus on one kind and keep up with the latest papers on the topic to some extent.

      That said, if you're sitting on the sidelines and not actively doing the job, it's feasible (though still hard) to keep up with most major innovations in DL as a whole. Only a handful of papers actually propose something that's both feasible and innovative.
      Aug 10, 2020
    • The vote count is likely explained by Blind being mostly engineers :)
      Aug 12, 2020
  • Walmart
    syntaxbugs

    Go to company page Walmart

    syntaxbugs
    MLE roles will continue to be popular for a while until the market realises that 99.9% of tech problems can be solved by traditional SWE roles.
    Aug 3, 2020 4
  • There's a lot of overlap.

    MLE: A person who can conceptualize, build and train ML models. Can put them into production etc..

    DS: Person with a more general statistical domain knowledge. Can probably prototype ML models if ML is the focus, but typically not capable of putting into production. But may be focused on other fields like time series forecasting, causal inference etc..

    Which is better? Depends. MLE will certainly pay more medium term. Long term, DS is little more business focused and reaching senior positions will be slightly easier. What's your objective?

    If you really only care about machine learning, be an MLE. As a data scientist you're going to be expected to solve all types of data problems. Nobody cares how you do it. As an MLE people expect a specific set of skills.
    Aug 12, 2020 0