Machine Learning Engineer vs Data Scientist

Waymo
RPWE37

Go to company page Waymo

RPWE37
Mar 11 5 Comments

I've been doing ML and writing code for several years now at various self driving companies. People ask me how to break into machine learning and data science, but to be honest I don't know what to tell them. For me I was interested in robotics, did some research and fell into this stuff.

One question I really don't know how to answer is the difference between a "Machine Learning Engineer" and a "Data Scientist". Are these just buzzwords for the same position? My gut mostly tells me that a DS usually helps make generic business decisions with data, while a MLE is more heavy on productionization. My usual response is to just look at job postings and see what fits them, regardless of title.

I really don't know. I write code, get it working know the car, and move on. If I need to I use ML algorithms. For those in more traditional MLE and DS roles, what do you see the difference as?

TC 380k

5

comments

Want to comment? LOG IN or SIGN UP
  1. TOP
  2. RECENT
  3. OLD
 TOP 5 Comments
  • You're right. Data science is more about crunching business insights from data and suggesting actions to stakeholders. Machine Learning Engineering focuses more on the software engineering aspects on data science with skillset such as Docker, Kubernetes, Airflow, Kubeflow, PyTorch/Tf, etc.

    There is a lot of overlap, but at the end of the day DS performance is evaluated by how many useful models/insights they presented, while ML performance is evaluated by production issues i.e. model retraining, latency, etc.
    Mar 11 0
  • New
    MaximusA

    New

    MaximusA
    MLE is more of a jack of trades, it can range from creation of data pipelines to building a distributed framework for model training, model management system and then deploying the model as a service. DS on the other hand will build the models from scratch. Think of MLE as an enabler for a DS
    Mar 11 0
  • PwC
    obiwan88

    Go to company page PwC

    obiwan88
    I just came to say thank you for using full meaning first before abbreviations. It irks me so much to see so many abbreviations with no explanation 👍
    Mar 11 1
  • Google
    oaDh23

    Go to company page Google

    oaDh23
    “Data Scientist” should have a graduate degree (2-years or more) in either Statistics, Biostatistics, Operations Research, or Econometrics; plus, at least 5 years experience as a Data Analyst. They know how to test and control for dozens of effects underneath the data to make the model more accurate (A ML Engineer will be happy with 70% accuracy and will have little idea how to improve it, a Statistician will push that same model past 95%). Classical Stats are vastly better than Modern-ML/AI models for nearly every application, except Computer Vision and NLP. Actual Data Scientists represent less than 1% of the modeling community, and most “Data Scientists” have no idea what an actual Data Scientist does.

    ML Engineer is someone using someone else’s off-the-shelf model, poorly, and not testing/controlling for dozens of effects that they’re unaware of. They pat themselves on the back a lot and don’t know what an actual Statistician does.

    YOE 15
    Mar 11 0