Airbnb Homes CTO Vanja Josifovski is urging executives in search of AI talent to look beyond top schools like MIT, Stanford University, and Carnegie Mellon University.

“I think you don’t have to go after a few graduates from the best universities in the U.S. to find the people that can do the job for you. That’s just not the reality out here,” Josifovski said. “In my prior experience, I have built programs sourced from community colleges, and I can tell you that in some cases I would hire people from that kind of nontraditional background, along with people from the top schools in the U.S., and over time, they will perform at a similar level. So I would encourage you to go and look at places other than those top schools.”

Managers should also consider that schools like community colleges may produce talent that, for a variety of reasons, including bias in society, did not attend an elite school.

A 2018 HackerRank survey of computer science students in countries around the world found that nearly 40% of respondents got their education from a combination of computer science school instruction and independent learning through things like online courses or YouTube.

Fervor to join the AI revolution has driven a number of businesses and organizations to seek out AI talent, but managers making hiring decisions continue to express a need for more people with job titles like machine learning engineer or data scientist.

Josifovski knows of what he speaks. Prior to joining Airbnb to lead engineering and data science for Airbnb’s Home division in March, he was named the first-ever CTO at Pinterest in 2017, where he developed AI-powered shopping personalization tech like automated ecommerce with Shop the Look or computer vision-enabled search tools like Complete the Look.

Josifovski spoke today with VentureBeat founder Matt Marshall at Transform 2019, an annual conference that explores artificial intelligence challenges and opportunities and major focus areas like computer vision and conversational AI.

Josifovski doled out a variety of tips during his talk, like the idea that enterprises should no longer attempt to build AI from scratch. In decades past, building an operation using data at scale required hiring talent for optimization, parameter tuning, inference, search, and others who can tie together pieces necessary to build AI from scratch, but public clouds mean that’s no longer required, he said.

“I don’t think there’s a real reason to build anything from scratch these days,” he said. “One of the most interesting things about this technology is that it commoditizes inside out. Usually technologies commoditize from the simplest part, and in AI it seems like the commoditization comes from the most complicated parts.”

As a person who spent most of his career in AI research, Josifovski said AI practitioners focused on optimization may be a better fit in major companies with AI research divisions like Facebook or Microsoft rather than companies like Pinterest or Airbnb, he said.