Here’s a quick summary of the first one! To watch the full interview, scroll down to the bottom of the article.
It’s constantly evolving. There’s a clear shift from machine learning and pure data science to a more holistic approach to roles. As companies collect more and more data, it’s inevitable to try to build predictive models. I think it’s a natural result of the age of data collection.
Success is achieved as a machine learning engineer from various backgrounds. Even though the role has been around for some time, it still feels new. While we look for diversity on our team, we try to hire people whose strengths will combine well with existing team members.
Educationally, solid understanding of computer science and math is standard. In addition, I’d say:
No one person is skilled at all of these equally. It’s incredibly difficult to find someone who would check all of these boxes immediately, but I suggest spending the time to develop these skills in a basic way, at a minimum. Then, figure out which areas really drive you and find a team that needs that energy. Because machine learning is changing all the time, it’s likely you’ll find a time that’s a fit for your skillset.
It’s a common mistake for companies to require machine learning degrees. Why? They’re relatively new, so there are few people with those specific credentials. Frankly, the demand for engineers with an ML degree doesn’t match the supply.
A lot of folks, myself included, come to machine learning engineering from quantitative fields. I have a PhD in physics. On our team, we have people with computer science, traditional software engineering, and mathematics backgrounds. They’ve all moved into an ML role well. You don’t need a specific degree to be successful, it seems very open to various experiences.
While pretty flexible, I’d say traditional software engineers who remember math concepts well probably have the easiest time and quickest path to success. Another extremely useful and valuable transferrable skill is the craft of software engineering. It can take a long time to develop that.
So, if you’re someone who has, you’ve got a big headstart. More ML engineer practitioners strengthen software development skills as they become more experienced, obviously. So, if you’re a software engineer, the pivot is a natural one.
Generally, they resemble interviews for software engineering roles. It commonly starts with a couple of technical interviews. You may meet with a cross-functional stakeholder, someone you’d likely work with on a project. This person might be from a department like Revenue, or Product Marketing and less technical.
The interview with the hiring manager may be toward the end. As for the technical portions, they’re commonly divided into software development and algorithms. The direct machine learning portion may use math concepts more directly.
There’s so many! Regarding modeling, dealing with textual data and natural language processing (NLP) are big. If you haven’t heard of the transformers revolution, it’s a new collection models incredibly efficient at comprehending language.
As for machine learning, MLOps is one to watch. At the crossroads of machine learning and DevOps, we’re seeing more and more roles in companies. Teams need someone who knows how to plan and execute deployment efficiently.
There’s also room for generalists. Machine learning skills are highly transferrable.
Related: In a survey of software engineers on Hired’s platform, they identified the hottest trend in tech as AI, Machine Learning, and Big Data, with 55.1% of respondents ranking it first. (Hired’s 2022 State of Software Engineers.)
Versatility and curiosity! Because the field is changing and growing quickly, keep learning! Learn the new techniques for modeling, technologies, the foundations – all of it. Don’t box yourself in by investing too much time in any one technology.