How have you seen AI and Machine Learning as a career path evolve over the years? What is fueling that growth?
Nico: "We've seen it evolving quite a bit even just in our team at Hired. It keeps evolving and changing every day.
We are seeing a lot of practitioners who are shifting from pure data science and machine learning to be more wholistic in their roles. On the job, we're pushing to deployment more frequently.
Companies are collecting an enormous amount of data these days. Far more than before.
And trying to build predictive models out of all this data is inevitable. I think we're seeing a natural byproduct of the data collection age."
What kind of educational background, technical knowledge, and soft skills are required to become a machine learning engineer?
Nico: "Luckily, in machine learning, you can see success with different backgrounds.
We value a lot of diversity in my own team and we tend to hire people to combine their strengths with others throughout the team. And even though the job has been around for a while, it still feels new.
You need solid foundations in math and in computer science, usually.
But beyond that you need also skills in statistical modeling, understanding probabilities, and statistics. You also need experience with software development and a sense for business.
You not only have to be great at managing data, but it's also helpful to learn about DevOps, which is a new trend.
You have to both learn how to develop models and deploy them to production smoothly.
Obviously, checking all these boxes is incredibly difficult. But I recommend spending time developing even just the basics of all of these skills. Then, find one or two areas you're passionate about and find a team that you can fit into.
Machine learning is changing all the time, so you'll likely find a team to complement your skillset."
Are there specific computer science degrees and certifications for Data Science and Machine Learning?
Nico: "The first bias companies have is to look for candidates with actual machine learning degrees.
They're still relatively new, but there are a few people out there with these credentials.
Obviously, the supply does not match the demand for engineers with an ML degree to their name.
There are lots of folks, like myself, who are coming from quantitative fields to machine learning. I have a PhD in physics.
But on our team, we've seen people join with backgrounds in mathematics, computer science, traditional software engineering, etc. They've all transitioned well into a machine learning role.
I would say it's very open to different background right now. No, you don't necessarily need a specific degree to succeed."
How can you pivot from different fields of engineering? What role is the easiest to pivot from?
Nico: "Well, again, it's pretty open, I would say.
Folks who see a lot of success are traditional software engineers who still remember a bit of math or can reactivate this knowledge from their college years.
And if you have some experience with the craft of software engineering because it takes years to develop that, it's extremely valuable and extremely useful in machine learning as well.
More and more machine learning engineer practitioners tend to develop software development skills as time goes by.
Even more so than pure machine learning status, I would say.
If you're a software engineer the transition can be pretty smooth."
What do machine learning interviews usually look like?
Nico: "To a certain extent, they resemble software engineering interviews.
It typically starts with one or two technical interviews. Then, you might have a less technical interview with someone from the project you'd be working on, someone from the revenue side.
And at the end of that there's there's usually an interview with the hiring manager.
The technical interviews are broken down into algorithms and software development on one side. On the other side, there's pure machine learning that may directly involve mathematical concepts..
What do you see as the biggest career growth opportunities in the machine learning ai space?
Nico: "Well the field is as vibrant as ever. There are so many possibilities these days.
On the modeling side, I think that natural language processing and dealing with textual data is booming.
We are at the beginning of the revolution called the transformers revolution.
It's a new family of models that are very efficient at understanding language.
You might have heard of GPT3 and the very impressive capabilities of this model—this is in terms of pure modeling.
I would say investing in natural language processing is probably a good call.
Then, in terms of machine learning, deployment and how to deploy these machine learning models, MLOps is growing.
MLOps is at the intersection of DevOps and machine learning. We see MLOps positions becoming the norm in several companies.
You can also be a generalist. Machine learning is applied to so many different fields.
You can stick to the foundations of machine learning and software engineering and hone your skills over time.
That is a very good idea because the skills that you learn are going to be highly transferable from one role to another.
What are the most important skills to continue developing as you progress through your machine learning career?
Nico: "It is a good idea to not invest too much in one specific technology that can be very appealing.
It is good to always keep learning because the field is evolving so fast. Keep learning the foundations, the new modeling techniques, the new technologies, but don't stick to just one.
Keep an open mind and stay versatile."
Ultimately, the best way to prepare for the machine learning interview is to get out there and practice. Here are some resources that could be helpful in your preparation:
👯♂️ Practice your behavioral and system design skills with our interview practice tool.
👨🎓 Take our complete Machine Learning Case Interview Prep Course
Good luck with your interview preparation journey!