Jonathan Woods
June 8, 2022

In a tight talent market, GoldSpot has succeeded at attracting some of Canada’s top young AI practitioners and building a team that keeps its R&D pipeline brimming with cutting-edge models and techniques. Remarkably, they’ve done this as a small, 80-person company in a sector that CTO Shawn Hood admits is “notoriously slow to adopt” new technologies – mining and mineral exploration.

Goldspot provides geologists with AI and data analytics-driven insights to make the search for new metals and minerals easier. Mineral exploration is a hit-and-miss endeavour that involves looking over new ground for clues and patterns gleaned from successful searches of the past, banking on the idea that finding the right mix of hints will lead to the discovery of major new deposits. It’s an economically risky, highly uncertain, and capital-intensive business. Technologies that can more reliably help to predict hits – like computer vision applied to historical drill cores or natural language processing to old exploration notes – can be a great asset.

But mining is also a very traditional industry, one not typically thought of as being at AI’s leading edge. So how has Goldspot been able to compete in the fierce contest for AI talent? Hood credits Goldspot’s academic-like culture and inspiring mission (“discovering the metals for the future!”), and one other important key: a very proactive approach to opportunities provided through their Vector Institute sponsorship.

“We’re mega-engaged,” Hood says. “And Vector channels have been very fruitful.”

Fruitful indeed. Goldspot initiated the hiring of five people on its AI team through Vector channels like the Digital Talent Hub, career fairs, and other events. It’s the payoff of being present. In Vector projects and collaborative sessions held over the years, Goldspot always seems to be in the room and at the table. They’ve become so deft at maximizing these opportunities to source talent that Goldspot data scientist Shervin Manzuri Shalmani – himself hired through a Vector career fair in 2020 – remarked that the company has “developed an appetite for hiring through Vector,” which has created “a network effect.” He says, “When you hire through Vector, you hire high-caliber people who know other very high-quality people.”

The success is also the result of Goldspot making itself an attractive place for talent to land. Employees that came through a Vector connection almost universally describe their shift from academia to Goldspot as “seamless,” praising a supportive culture that, according to Shalmani, encourages them to “continue to probe this AI behemoth, and find nooks and crannies where improvements can be made.”

This is very energizing for the ambitious AI scientist. Sam Sattarzadeh, a machine learning engineer who connected with Goldspot through Vector’s Digital Talent Hub, says, “It highly motivates the whole data science team to go through all their academic and engineering tasks, to make the products, when they realize they can also work on a high-level paper.” The motivation is evident. Two Goldspot data scientists brought on through Vector channels have authored a workshop paper for this year’s Conference on Computer Vision and Pattern Recognition (CVPR), the world’s leading computer vision conference.

Another appeal is the opportunity to make a real impact in the company and on the industry. “The market for AI talent is a very competitive space,” says Hood. “You can work at a large company, and it’s going to be a fight to make a name for yourself and be top talent. Or you can work in mining, and you can really shine in the industry.”

Shalmani agrees: “You get to own your contributions and say that you developed an algorithm or AI pipeline. You sort of start owning it, and you can say in your mind that what I’m doing is making an effect in the mining industry.”

Retaining highly-driven talent requires staying at the cutting edge. That’s supported, in part, by Goldspot’s engagement in Vector industry projects that bring sponsors together to explore the application of leading AI research to industry use cases. Goldspot participated in Vector’s Natural Language Processing project, which can help automate what Hood calls “boring, rote, repetitive tasks,” like going through decades-worth of historical paper documents to spot notes about specific minerals that might help identify deposits elsewhere. They’ve also joined Vector’s new Synthetic Data Project, which focuses on creating AI-generated data that can be used to train models when well-organized, real-world data is limited – a major problem in geology.

Hood is very clear that zealous collaboration with Vector is a deliberate decision. “What I was looking for was a group or institution that’s going to be the focal point for people across academic institutions, across companies,” Hood says. “When Vector appeared, I thought, ‘Perfect, yes, this is the group we’re going to work with.’”

He continues: “And we’re going to be engaged if they need something from us, because you have to invest to receive the full benefit. Being an active and valued contributor is one of the best aspects of Vector involvement.” For Goldspot, this involvement has led to an invaluable source of homegrown AI talent to support the discovery of metals for the future.