4 ms·

I'm not sure how well this generalizes. I'm a Machine Learning Engineer, and that seems to check all those boxes. It is niche, it has low offer, and crazy high

I'm not sure how well this generalizes. I'm a Machine Learning Engineer, and that seems to check all those boxes. It is niche, it has low offer, and crazy high demand. Sure, salaries are high, but the interview process isn't easier, in fact is way harder. As an MLE I need to excel both as a software engineer (yes, that includes leetcode), and in MLE/science/data science. For example, the last position I entertained included a phone screen and then 4 rounds: behavioral, coding, machine learning expertise, and a science round, where I was given a problem, had to find 2 papers and make a presentation explaining my approach to solve the problem using the ideas from the papers. At this rate I'm considering leaving Machine Learning altogether, since the barrier to entry is too high for the people already in.

6gvONxR4sf7o 4y ago
MLE is different because there’s a ton of people who want to do it, and apply without the skillset.
t_mann 4y ago
What exactly do you mean by low offer? I mean, my idea of a niche isn't something that is among the top most popular courses on every MOOC platform. What's a barrier to entry for people already in? You mean for switching employers? What do recommend to get in?
- angarg12 4y ago
  I mean offer as in offer and demand. There isn't many people with the engineering skills, ML expertise and on-the-job experience. The barrier for entry means that interviewing for the job, even with experience, is very difficult. As others said maybe the different is there is huge demand. Still I find it paradoxical. You'd think having experience in a field with high demand and low offer would be a pretty good deal. And yet it would be easier for me to find a job as a generalist than as an specialist. My advice to get in is based on my experience so YMMV. I'd say the easiest way is to transfer internally to an ML position. That's how I got here. If you are a solid engineer, teams are happy to fill in your ML knowledge gaps. Anecdotically most people I interview got to ML the same way. If your company doesn't have official ML Engineer roles or ML projects, you can position yourself to jump in when the opportunity arises, or even pitch a project yourself (leadership seems to be suckers for ML-for-anything). The second easiest way is as a student/new grad/junior. Expectations for juniors are lower so it's always easier to get in. The ML round of the interview is mostly theoretical so students aren't at a disadvantage.

6gvONxR4sf7o 4y ago

MLE is different because there’s a ton of people who want to do it, and apply without the skillset.

t_mann 4y ago

What exactly do you mean by low offer? I mean, my idea of a niche isn't something that is among the top most popular courses on every MOOC platform. What's a barrier to entry for people already in? You mean for switching employers? What do recommend to get in?

angarg12 4y ago
I mean offer as in offer and demand. There isn't many people with the engineering skills, ML expertise and on-the-job experience. The barrier for entry means that interviewing for the job, even with experience, is very difficult. As others said maybe the different is there is huge demand. Still I find it paradoxical. You'd think having experience in a field with high demand and low offer would be a pretty good deal. And yet it would be easier for me to find a job as a generalist than as an specialist. My advice to get in is based on my experience so YMMV. I'd say the easiest way is to transfer internally to an ML position. That's how I got here. If you are a solid engineer, teams are happy to fill in your ML knowledge gaps. Anecdotically most people I interview got to ML the same way. If your company doesn't have official ML Engineer roles or ML projects, you can position yourself to jump in when the opportunity arises, or even pitch a project yourself (leadership seems to be suckers for ML-for-anything). The second easiest way is as a student/new grad/junior. Expectations for juniors are lower so it's always easier to get in. The ML round of the interview is mostly theoretical so students aren't at a disadvantage.