Capital One CTO on the drive to expand talent in Canada

Capital One Canada is building a lasting engineering presence, and its Chief Technology Officer, Damjan Pelemis, said the effort is helping the bank attract and retain Canadian talent.

He says the talent challenge has only intensified since the post-COVID Great Resignation: skilled people are hard to find and even harder to keep in Canada. To attract them, Pelemis argues, employers need to offer purpose and growth opportunities that help people succeed within the company and beyond.

That approach underpins Capital One's international talent expansion, giving engineers and product managers in Canada the chance to work on products and services in Canada and across the border.

"While there is transfer and we encourage it, it's truly a rounding error on the total number. I credit that largely to people liking it in Canada and liking to not just work here but live here. This international expansion gives them the best of both worlds. Work for a much bigger market, Canada plus U.S., without having to leave the [Greater Toronto Area (GTA)]," said Pelemis.

Capital One recruits new graduates from across the country, from British Columbia to the Maritimes, drawing on top-tier universities, Pelemis said. Mid-career hiring is largely concentrated in the GTA, while senior hiring is Canada-wide because those candidates are generally more willing to move to the city.

He said graduate hires for the international expansion are typically software engineers who enter through the company's Technology Development Program. Recruits start in undifferentiated roles and specialise in mobile, backend or frontend development over time rather than on day one. Key feeder schools include the University of Waterloo and the University of Toronto, with Queen's University also a strong contender.

The company has shifted what it looks for in new graduates, Pelemis said.

"I think the scale has shifted materially. Both with what the schools teach with respect to skills, or when we look at the entrance testing and qualification. That goes to coding familiarity, to coding and computer science basics and structures, not for the sake of familiarity, but for the sake of informing judgement and taste."

The company puts a lot of weight on behavioural interviewing, looking for people with the skills and a propensity to show them in a team, communicating and working across functions. 

Human-first expertise is still top of mind. Frontier AI models help with only a small part of the problem, he said. Implementation costs have fallen sharply, but knowing what to build remains difficult, and asking a model what to code returns many answers. He said that when the company launched Quicksilver World Elite Mastercard, a cashback card with no monthly fee for the Canadian market, the frontier models it works with had no knowledge of specific features, requirements, market positioning, benefits, eligibility, or internal implementation context.

"That context and choices had to come from our teams. The models could assist with the build once given the right information, but human expertise was required to determine the product choices and guide the implementation," he added.

AI skills aren't new, he said, but they emphasise the value of judgement. Five years ago, the company had a smaller machine learning presence, and it wouldn't hire out-of-school machine learning talent today.

"In the new world, you can create 1000 launches in one day with one click. But discerning which one is valuable and which one do you want to follow now becomes a lot more valuable," said Pelemis. " So refining that skill, discernment, judgement - not just for the product, but for tech as well. What does a good tech product look like? How does it manifest its health and resilience? Those aspects becomes so much more important."