The title is a trick question, of course. Much like Data Scientist before it, the title Machine Learning Engineer is developing into a trend in the job market for people in our profession, but there is no consensus about the meaning of the title or the functions and skills it should encompass. I imagine new entrants into the job market in DS/ML find this maddening to decipher. (Even experienced people do!) So, let’s talk about what it might mean depending on who’s doing the talking.
As I was discussing this with a friend the other day, I phrased it as MACHINE LEARNING engineer or machine learning ENGINEER. Basically, from what I have seen, there are roles and expectations in some camps under the title that either:
- A. expect extensive software engineering skills with a dollop of experience or at least familiarity with ML, or
- B. there are expectations of intensive ML experience, often including deep learning or generative AI, and they would like it if you can write a function when it’s called for.
The former group probably would have just been “Software Engineers” in previous years, while the latter would have fallen comfortably under “Data Scientists” back in the day when I started my career (although generative AI was certainly not part of the game back then).
This reflects an interesting pattern in the development of our profession more broadly. We have never been good at breaking up the roles in our field into subcategories that clearly delineate the skill set (or the responsibilities) of the roles. It is a fast moving, constantly changing young field, so this is not shocking! This was always true of the title Data Scientist, which was essentially a delineator for “something more technically skilled than a Data Analyst” for a long time. Some folks referred to Data Scientists as the people who could handle unstructured or disorganized data, and that’s gone away as a defining factor from what I can see.
I strongly suspect the growth of MLE is because people hiring SWE type folks were annoyed to not be getting candidates who knew their way around an ML model, while the people hiring the Data Scientists were getting analytics specialists when they needed modelers with ML skillsets. They intersected from each direction, forming a new title that has internal disagreement about the importance of each skill set. So now we’ve got a new division in the space to think about.
While this subdivision of the field is probably very natural, as a response to this sort of difficulty, I want to make a point about what this means for candidates and the field. Anytime a new split happens and the career path has a new possible divergence, there is status and privilege assigned to the two routes, most often detectable by the salaries on offer for each direction. Now that the field of Data Science is becoming formalized with more education opportunities and such, people have easier pathways into the career. This includes people who are disadvantaged or marginalized in broader society. I believe we are at risk of a pink-collaring effect of Data Scientists.
(In short, the pink-collar effect is when women in particular become a larger proportion of workers in a given field and the salaries and social status for the roles where they predominate are depressed systematically because of it. Veterinary science is a common example. It goes the other way too, as women were predominant among computer programmers in the 1960s and early 1970s, and when men became more represented in that field, their salaries and prestige went up.)
Is this actually happening? I don’t truly know. I only see anecdotal evidence from industry reports from places like Harnham and Burtch Works, as well as browsing the job postings on places like LinkedIn, that make this seem like a salary divergence is starting to occur between Data Scientist and ML Engineer. I certainly meet many more young women, POC, and people of different gender identities and sexual orientation in Data Scientist roles than I did even five years ago.
I hope very much that researchers might be able to find out if this salary shift is statistically significant, and if so, if it corresponds with changes to the demographics of workers the way I suspect it might.
At any rate, the challenge for those hiring in the field is to not allow the more prestigious, more “technical” hires (eg, now Machine Learning Engineers) to become predominated by men and those with social advantage, and by corollary for the Data Scientist hires to become a lower status variant that others in the field get shunted into regardless of ability. Pay the roles what they are worth to your business, of course—but don’t let that affect the demographics of people you consider or envision in each role. That’s the least we can all do at this stage of the ever-evolving game.
You can find more of my work at www.stephaniekirmer.com.