Machine-learning specialists are among the most in-demand talent in tech right now, with major firms such as Snap, Zoom, and Microsoft advertising for engineers.
Machine learning is a branch of artificial intelligence, with engineers usually designing programs and algorithms that can learn. Features like facial recognition, Facebook’s news-feed rankings, and Apple’s voice assistant Siri feature machine learning. As applications for artificial intelligence grow in technology, healthcare, finance, and other industries, demand for specialists has exploded.
Machine learning engineer was ranked the second-fastest-growing job in the UK, according to LinkedIn, and the fourth-fastest-growing in the US.
Base salaries for the role in the US range between $72,600 and $170,000, according to LinkedIn. Major firms will pay even more. According to disclosure data on foreign labor hires to the US analyzed by Insider, a machine-learning engineer at Apple was paid $250,000 in base salary in 2021. UK data is harder to come by, but at the higher end salaries come in about 89,000 pounds (about $121,000) including bonuses, according to Payscale.
To understand what it takes to break into the field, Insider spoke with researchers at the Alphabet-owned research unit DeepMind, whose groundbreaking AlphaGo program beat human champions at the board game Go and whose AlphaFold program can predict the structure of proteins. We also spoke to Tractable, a British company that uses computer vision to assess car damage.
Qualifications required for machine-learning jobs range from a minimum of a bachelor’s degree in a technical field such as computer science to a master’s degree or doctorate in computer science, mathematics, physics, or other highly quantitative subjects.
Luis C. Cobo, a senior staff research engineer lead at DeepMind, told Insider that strong software-engineering abilities “are necessary but not enough” on their own.
“Candidates must have hands-on experience applying the scientific method to the AI/ML context, making hypotheses, running experiments, and drawing well-founded conclusions from the results.”
Tractable has a degree of flexibility when it comes to an applicant’s background.
“What we care most about are their technical capabilities and technical excellence,” the company’s chief technology officer Razvan Ranca said. “So we have a wide range of backgrounds.”
“We have quite a few physicists,” he added. “We have neuroscientists, chemists, all who have retrained in ML over the years.”
Although some programming experience is necessary, specific skills can be taught to qualified applicants. A candidate’s mindset is also important.
“Not every material academic researcher has the inclination or the ability to see something through from idea all the way to running on, you know, millions of households,” Ranca said.
And Chris Gamble, also a senior staff research engineer at DeepMind, told Insider he hires people “from all types of academic backgrounds.”
“Indeed you don’t even need to have a degree in an AI- or computer-science-related field to be successful,” he said. “We look at the potential to succeed, demonstrated ability in AI/ML, ability to code, and an entrepreneurial drive.”
If you are selected to interview for a job at DeepMind, you can expect at least three interview rounds, Gamble said.
“First, an initial conversation with the team lead and our recruitment team; second, a series of technical interviews focusing on coding ability and applied AI knowledge; and third, a couple of interviews to assess which team makes most sense for you to join.”
Tractable’s Ranca also emphasized an entrepreneurial attitude — at least for engineers who want a role at a growing firm. “We look for people who want to really take ownership and really drive results,” he said.
At a more senior level: “We look for people who have shown that they can take something from an idea all the way to an actual system that functions in production and delivers value,” he said.
There is a general shortage of deep-learning skills at the moment. For candidates in related industries who want to move into machine learning, entry-level roles or internships are available.
“Some of our best and fastest-growing researchers started with an internship because they were basically switching from a completely different field,” Ranca said.
Meanwhile, Gamble told Insider that while many DeepMind employees start as junior engineers, career progression can “take many paths.”
“Generally we focus on two potential paths: as an individual contributor or as a people manager,” he said. “The separation of these paths isn’t always distinct, and many people straddle both individual contributor and management roles.”
Gamble added that the three main factors the company looks at when assessing an employee’s path are: “Contribution to DeepMind’s mission, collaboration and leadership, and knowledge development.”