Skills related to artificial intelligence, machine learning, and data science remain in high demand, with companies scrambling to fill key roles in their pursuit of deep data insights to drive decision-making and to make the most of the promise of AI.
For those companies where AI is the key to the future, the skills shortage is of particular concern. CRM SaaS provider Salesforce definitely falls into that category. The company’s Einstein platform is one of the most common ways enterprises deploy AI tools, as for most companies, AI built into platforms they already use is a key means for leveraging this emerging technology.
According to IBM’s global AI adoption data, the single biggest barrier to AI adoption is a lack of skills, even as 43% of companies say they’re accelerating their rollout of AI as a result of the pandemic. For Salesforce, which this past year hit an earnings records, with growth of 24% to $21 billion in total sales, in some part thanks to its AI-powered Einstein platform, getting creative in pursuing AI and data science talent has been essential.
Einstein predictions went up from 1 billion a day in 2019 to 80 billion a day as of this past November, a feat that would not have been possible without a multiprong approach to finding and upskilling talent, says Marco Casalaina, senior vice president of product management and general manager of Einstein at Salesforce, one that includes tapping a range of talent pools beyond blue-chip universities.
Finding talent in nontraditional places
For Salesforce, supporting the growth of its AI-related platforms and services is not just about mining the usual corporate recruitment channels. While most companies seek out relationships with local or traditionally tech-rich universities, Salesforce partners with more than 700 colleges and universities around the world.
And in doing so, Salesforce is finding that it’s getting easier to find graduates who have the necessary skills, Casalaina says. That’s not just thanks to the breadth of effort, but shifts in the coursework required at many universities these days.
“A lot of the folks that come to Salesforce as developers nowadays do have some data science background,” he says. “It’s now part of the core curriculum of computer science. They’re able to do this more or less natively. That’s not such a bad thing.”
Salesforce also runs a program in cooperation with Year Up, a non-profit working to redefine and diversify the IT talent pipeline, having helped more than 34,000 disadvantaged youth find opportunities in finance and technology.
“The Year Up program brings youth from underprivileged backgrounds as interns here, and a lot of them become employees,” says Casalaina.
For companies seeking expertise from the other end of the experience spectrum, Casalaina recommends looking at AI research papers at sites such as arXiv.org to find talented experts, and to mine employees’ networks.
Many of his team members have PhDs in mathematics or data science, he says — but not every team member needs that deep a level of AI expertise.
“Data science is 20% of AI,” he says. “A lot of the AI is getting the data in the right shape and making it work with human intuition.”
The work requires familiarity with the Salesforce platform and an understanding of how customers use it, he says. “Machine learning is part of it, but it’s not all of it,” he says.
So another nontraditional recruitment strategy Salesforce has undertaken is to look for people who are not currently in AI-related roles who are interested in making the move. According to an analysis conducted by LinkedIn earlier this year, half of employees who have moved into data science and artificial intelligence jobs have come from unrelated fields.
At Salesforce, that includes internally recruiting nontechnical staff. “Lots of people in our company want to be part of the AI group, and it’s beneficial to have folks from around the company come join us,” says Casalaina.
Employees who move laterally within a company already understand the company’s culture and its product lines.
“One of the guys on my team came from sales,” says Casalaina. “Brad was a salesperson for Salesforce. He sold our sales products, including some of the AI ones. Before that, he was doing sales for a pharmaceutical company.”
Now, he’s a product manager for Einstein. “He manages a couple of teams of developers, works with engineers, sets the requirements, talks to the customers,” says Casalaina. “Today, for example, Brad and I were talking about user research.”
To perform his job, Brad doesn’t need to be steeped in Python, a programming language popular with data scientists.
“He’s welcome to do that — it’s always good to know the deeper layers of machine learning,” says Casalaina. “But because of the way the Einstein platform is structured, he doesn’t have to go all the way down there. He needs to understand machine learning, and he certainly does. But he doesn’t necessarily need to make a decision about whether to use a neural net or a random forest. The Einstein platform can automate a lot of that.”
When employees move over to the AI team from other areas of Salesforce, they typically start out by taking Salesforce’s internal AI for product managers course or other internal certifications Salesforce has to offer.
“A lot of folks start that way,” he says. “Some folks also get certifications externally from places like Coursera.”
Accessing the Trailhead
The major platform for Salesforce training is its Trailhead platform, available both to internal employees and the general public. “We have a whole Einstein system on Trailhead and a ton of content,” Casalaina says.
That content includes more than 200 learning modules and 15 projects. For example, users can build a cat rescue app that recognizes different cat breeds.
“I’ve actually written a number of them myself,” he says. “With the hands-on ones, it actually provisions you a section of Salesforce where we have data preloaded and you can go in and try an Einstein feature and we guide you through the process. It uses data that’s made to work with machine learning, very specific data sets that are made to be realistic.”
In addition to creating content for the platform, Casalaina says he’s also taking some of the courses himself.
All the content is free to the public, he says. Companies seeking to leverage Trailhead to train up their own staff can sign up for myTrailhead for Employees, for $25 per user, per month.