Everyone’s Seeking AI Engineers — Here’s What They Want
There’s no doubt: Machine learning and artificial intelligence are the hot specialties in IT right now — but filling those jobs is proving to be tough.
In a September Gartner survey of over 400 global IT organizations, 64% of IT executives said that a lack of skilled talent was the biggest barrier to adoption of emerging technologies, compared with 4% the previous year.
Companies are looking for employees with specific training, skills and personal traits to fill positions — STEM degrees, credentials specific to AI and machine learning, practical hands-on experience, and certain soft skills are all considered when deciding whether to hire a candidate.
When Demand Exceeds Supply
The current push to find AI developers and engineers makes the shortage of candidates undeniable, and makes hiring particularly grueling.
In a November survey of over 2,500 human resources and engineering personnel by HackerEarth, a software company that helps organizations with their technical hiring needs, 30% of respondents said they’re expecting to hire more than 100 developers in the coming year.
With goals that ambitious, a significant portion of those hiring managers are so much in need for talent that they’re willing to compromise their standards. Nearly 35% of engineering managers said they would compromise on candidate quality to fill an opening quicker — and nearly 24% of HR managers said the same.
According to the survey, AI and ML experts are in high demand this year, with demand exceeding supply.
“What we have today is a rich tapestry of interrelated jobs or personas that all go into creating a data science or AI outcome in the enterprise.”
—Bradley Shimmin, chief analyst for AI platforms, analytics and data management, Omdia
“It’s a candidate’s market out there,” said Vishwastam Shukla, chief technology officer for HackerEarth.
With companies from all industries looking to hire, larger organizations have the advantage of being able to offer bigger salaries and plusher benefits, he acknowledged. But he’s seeing smaller employers and fast-growing startups put up a good fight for candidates.
One of the most popular tactics, Shukla said, “is to actually inculcate a culture of learning and development within the organization.”
How AI Jobs Have Evolved
The AI positions companies are looking to fill have become narrower and more specialized.
Job requirements vary wildly depending on a company’s size, how mature they are, their data infrastructure and what kind of projects they’re working on, said Bradley Shimmin, chief analyst for AI platforms, analytics and data management at global analyst firm Omdia.
“Five years ago, ‘data scientist’ was considered the hottest job on the planet, and we were talking about data scientists as unicorns in that they possessed a number of very specific skills — mathematical, statistical, business and communication,” he said.
Companies realized early on that they couldn’t operationalize with just a few jack-of-all-trades data scientists.
“Trying to scale with them was impossible financially and that, coupled with the creation of MLOps platforms, really spawned a diversification for the job role and a slicing off of aspects of that job,” said Shimmin. “What we have today is a rich tapestry of interrelated jobs or personas that all go into creating a data science or AI outcome in the enterprise.”
What Skills Does an AI Engineer Need?
The job titles of AI and ML engineers and developers cover a wide variety of tasks and responsibilities, but there’s a lot of overlap.
A necessary background for a potential employee starts with programming experience and a college degree.
Companies specifically look for:
- At least a bachelor’s or master’s degree in computer science, data science, machine learning, artificial intelligence or a related field.
- Programming experience with Python, C++, Java or similar languages.
- Experience with SQL to interact with data.
- For ML engineers, a knowledge of data tools like H20 and TensorFlow.
And companies are hiring anywhere from basic entry-level positions to more advanced roles.
“We’re just hiring at all levels,” said Valerie Junger, chief people officer at Quantcast, a technology company that focuses on AI-driven real-time advertising.
Machine learning engineers have to be fluent in Java, C++, Python, or similar development languages, and need anything from a master’s degree to a Ph.D., depending on the role, she said.
Just having a general computer science degree isn’t enough — recruiters look for an applicant who’s taken specific courses in AI and ML.
“In the past, I would check that applicants had a math or STEM background only,” said Rosaria Silipo, head of data science evangelism at KNIME, a data-analytics platform company. “Now, with the proliferation of college programs and online courses, I check if they have any credentials specific to machine learning or data science.”
The requirements for a machine learning engineer have changed, said Omdia’s Shimmin: “All the platform players, Microsoft, Google, Amazon, and others are setting up certification programs.
“You don’t need to have a Ph.D. — you can take whatever time it takes to prove certification as a machine learning engineer, or a data learning engineer, or as a machine learning specialist, and you can put that to work,” he said. “You can have a bachelor’s or a master’s and still get into this area.”
Pursuing specific credentials can lead to better jobs, or to a pay bump in a current position.
According to an October survey of over 3,000 data and AI professionals by learning company O’Reilly, 64% said they took part in training or obtained skills to build their professional skills, and 61% participated in training or earned certifications to get a salary increase or promotion.
And over a third of those polled dedicated more than 100 hours to training. Those survey participants reported an average salary increase of $11,000.
Gaining Experience
Entering competitions or hackathons can make a person stand out in a pool of prospective AI/ML candidates who have similar degrees and credentials.
For a candidate, entering hackathons helps potential employees connect with companies and learn a lot about how an organization works.
“In the past, I would check that applicants had a math or STEM background only. Now, with the proliferation of college programs and online courses, I check if they have any credentials specific to machine learning or data science.”
—Rosaria Silipo, head of data science evangelism, KNIME
For an organization looking to hire a lot of people quickly, hackathons can provide a bounty of leads.
“Hackathons let you create this warm pool of talent, because a lot of times when you actually go out to hire in the market, you may not be able to source the right kind of candidates with the right skill sets at a short notice,” said HackerEarth’s Shukla.
For entry-level candidates, one of the most direct ways to learn how a company operates is through interning — and a company can see if they’re a good fit.
“We try to bring on interns who we can get to know before they graduate and they get to experience our culture beforehand,” said David Karandish, founder and CEO at enterprise AI software-as-a-service company Capacity.
“We really lead with ‘Hey, here’s the type of work you’re going to do here.’ And we like people who are excited about the work that they do.”
No Jerks Allowed: Soft Skills and AI Jobs
In the DevOps era, teams need to be increasingly cross-functional as businesses and data-driven product development come together. Good communication and collaboration skills are considered as important as a degree or a certification.
AI professionals need to explain complex topics — often across multiple time zones, in a remote work setting, and be understood by a wide variety of people with various levels of technical knowledge.
No one person is ever going to know how every single thing works, noted Karandish. So organizations need people who can collaborate and coordinate together, and know when to ask for help or to bring up an important issue.
“It’s knowing when to ask, are we going down the right path or not, or is there a different approach?”
And attitude goes hand-in-hand with collaboration.
“Nobody wants to be working with a jerk,” he said. “They tend to not be collaborative and tend to take credit when credit isn’t due. So we’d like people with a high-talent-to-low-ego ratio in general.”
A wide variety of companies are hiring, and an AI professional needs to understand the specific issues they’re trying to solve for their employer.
“They need to have the proper domain knowledge to be able to provide precise recommendations and critically evaluate different work models,” said Kamyar Shah, CEO at World Consulting Group.
“To design self-running software for businesses and customers, they need to understand both the company and the issues their designs solve for that company,” he said.
Problem-solving is another highly valued skill — not just understanding what a problem is, but being able to come up with new solutions.
“A big aspect of ML and AI is creating playbooks that have not been built before,” said Wilson Pang, CTO of data company Appen. “A developer needs to have the ability to try new techniques, test and learn, and continually grow through keeping up with industry trends.”
