I am thinking about moving to FB, currently an Applied Scientist II at Amazon. I have seen a lot of job families at FB, unsure what’s the difference, or which one to apply to. For example: ML engineer Data Scientist Research Data Scientist Research Scientist Can anyone comment on: (1) Which one pays more/less (2) Which one is more advanced in ML? (3) Is any of these “less than” compared to the others? In Amazon, it’s pretty clear: DS < RS < AS, but in FB I think it depends a lot on the team. I am interested in improving my ML expertise, perhaps with some production work, but not entirely focused on it. Thanks! #machinelearning #machinelearningengineer
DS<RDS<RS=MLE
Does this apply to salaries as well? Or is MLE paid higher to compensate for potentially boring work? Does RS usually require PhD? I am guessing MLE doesn’t
Salaries as well, MLE*0.6=DS (same base, stocks different). None requires PhD explicitly, you just have to pass interviews. RS and MLE are the same thing at Facebook. You may not be able to match with certain teams without the necessary background, if they are looking for extensive research expertise. MLE is roughly equivalent to AS at Amazon.
ML Engineer is called Software Engineer, Machine Learning at Facebook. Is really Applied Scientist more than Research Scientist at Amazon?
Yes. Applied scientist here as well. At Amazon an AS is a superset of RS+SDE, that is, expected to be able to write production level code.
3 AS from my team left Amazon for FB. Two have a software engineering role. The third one is software engineer, machine learning specialist.
For the one who is SW-ML specialist, do you have any idea how much ML he does? I hear is common for those positions to end up having only 5-10% ML
Not sure. His expertise is RL. He has a postdoc from MIT.
At Facebook by and large RS, ML engineer or any other engineer is primarily a software engineer. You build the pipes, design the system and code them up. Most ML work is platformed, so very little engineers need to do such work. Your primary task is software engineering.
What about the people who publish a lot? What is their title vs these kind of RS?
I know people who publish 3+ papers in iCml neurips etc at FB each year. I’m sure they’re not doing the kind of work you are describing
No idea
Ds positions can’t match your as TC so I was told. The recruiter recommended interviewing for SDE- ML.
You mean “Research DS” at FB couldn’t match AS at Amazon?
I meant DS- products. I am not familiar with research DS
Following
There are some AS on my team at Microsoft who came from Amazon. There are others who left my team and went to FB as SWE ML / MLE titles. I am surprised though, since on my team we definitely do more ML modeling rather than writing production code. Even though AS on my team means SWE + ML experience, oftentimes it seems like there's way less SWE work and more ML work. Since many models never make it to prod. I guess in the FB interviews, if you study enough LC and ML system design, then you're all set?
What FB level do your teammate at Microsoft leave for and get from Fb?
Unsure
Anyone know whether PhD is required for research at FB?
For the tiny number of teams that actually do research yes. Otherwise it doesn't matter, cause we don't do research
Thank you and good to know. FAIR jobs sound cool, but so does building real productionized models.
Following.