How two small Canadian companies compete for AI talent

At a time when the skills of artificial intelligence researchers are very much in demand by deep-pocketed tech companies willing to pay hundreds of thousands of dollars, nimble startups offer an alternative for the field's rising stars.

Smaller teams, a sharper focus, and the draw of startup life help attract potential hires

There is great demand for researchers to work on artificial intelligence, both from big tech firms and small startups. (Evan Mitsui/CBC)

Maluuba is an increasingly rare beast in the burgeoning world of artificial intelligence. Founded in 2012, the Waterloo-headquartered startup has yet to be acquired by one of the many massive tech companies investing heavily in AI. Nor has Maluuba been raided for its talent, which is very much in demand.

Rather, the small artificial intelligence startup is now home to more than 50 employees and growing — and for prospective hires weighing opportunities with larger firms, co-founder Mohamed Musbah thinks he has a pretty good pitch.

Maluuba is trying to train computers to understand language, whether in conversation with a human, or when reading a document or other source of text.

We're just focusing on one area, and we're trying to push that really, really well- Mohamed Musbah, Maluuba

"If you actually solve language," Musbah says, "you've taught machines how to think, reason, and communicate." 

To some extent, nearly every big technology company is working to solve this problem — albeit, in the service of larger product development efforts in areas that range from advertising to product recommendations. That's exciting for some researchers and engineers, but others might prefer a singular focus on a niche area. 

Musbah believes that focus is Maluuba's appeal to young talent.

"We can't compete with Google or Facebook or Microsoft across every angle,"  Musbah says. "We're just focusing on one area, and we're trying to push that really, really well."

A 'land grab' for talent

In recent years, artificial intelligence has been hailed as the future by many technology companies for its potential to make their products and services smarter and more useful. Google, Uber, and now Apple have been using AI and computer vision to train self-driving cars, while companies like Facebook and Twitter have formed AI Labs to make sense of the ever-growing troves of data created and uploaded by their respective users.

Others, like chipmakers Intel and IBM — which have lost much of their influence over the technology industry over the past decade — see artificial intelligence as a potential path to relevance once again.

Uber released this image of its first driverless car. Google, Uber and Apple have been using AI and computer vision to train self-driving cars (Uber)

As a result, the entire tech industry is engaged in what Andy Mauro, the CEO and co-founder of Montreal-based AI company Automat, refers to as a "land grab" for AI talent. Big-name players have paid hundreds of thousands of dollars in yearly compensation to poach star researchers and engineers. In some cases, big players acquire smaller firms, and their talent, wholesale.

"The talent fight is on for sure," said Mauro, whose company has just nine employees — but he believes that the industry is also starting to cleave into two camps.

"There are people that want to work at big companies and have that stability. And there are people that want to be pirates and go work at startups," he said.

Having an impact

At Automat, Mauro and his team are developing software so that brands and marketers can easily launch conversationally savvy chat bots powered by artificial intelligence. 

Talking to computers is one of the hardest AI problems out there- Andy Mauro, Automat

"Talking to computers is one of the hardest AI problems out there, and it has real meaningful impact in terms of how we interface with technology day to day," Mauro said.

The pitch to potential hires at both Maluuba and Automat is that their teams are much smaller, and closer knit, than larger firms — and that everyone is laser focused on solving the same problems. And where larger companies can sometimes be secretive about their efforts, the smaller companies tout the freedom their employees have to collaborate and openly publish their research.

Mauro and Musbah have both found that the chance for employees to have a real impact at their respective companies can be another big motivator, too. They say that, because startups are often more nimble than big tech firms, employees can see the fruits of their efforts in actual products and services much faster than at a big tech company like Facebook or Google. 

"When I say 'you are an AI researcher, and you work at a 10-person company,' you are literally responsible for the success or failure of that company and all the other nine people you work with are as well," Mauro says.

Montreal leading the way

Computer science professor Yoshua Bengio, shown at his home, has helped lure tech talent to Montreal and recently launched an incubator for AI talent. (Graham Hughes/Canadian Press)

This time last year, some expressed concern about a brain drain of AI talent from Canadian companies and schools, after a series of hires and acquisitions by Silicon Valley companies and other tech hubs in the U.S.

But there has been optimism in recent months, with a focus on one city in particular: Montreal. Though the city has long been a hub for artificial intelligence research — thanks to the contributions of longtime deep learning scientist Yoshua Bengio, a professor at Université de Montréal — a series of high-profile announcements have only further cemented its reputation as a leading tech hub.

In September, the Canadian government announced $213 million in funding for Université de Montréal, HEC Montréal, Polytechnique Montréal and McGill University to further artificial intelligence research in the city. The following month, Bengio announced an incubator for AI talent called Element AI.

And in November, Google said it would give $4.5 million to the Montreal Institute for Learning Algorithms for research initiatives — though not without announcing it was opening a Montreal AI research group of its own.

Maluuba also opened an office in the city around this time last year — and while some graduates may be drawn to the promise of Google's new AI lab, Musbah thinks that the challenge of working solely on building systems that can learn to read like humans will be its own draw.

"Our angle, or our hook, is around that idea," Musbah said.