These days, everyone is asking questions about the implications of machine learning tech on our society. With our high concentration of machine learning talent in cities like Toronto, Ottawa, and Montreal, Canada seems to be a growing part of the plan for global tech companies eager to dominate the space. Microsoft recently acquired Waterloo-based machine learning startup Maluuba, while in Montreal, Element AI was the first organization to receive investment from its venture arm, and Google made its own investment in a major Montreal-based machine learning hub.
When Kerry Liu started Rubikloud over three years ago, he saw the potential for machine learning to disrupt almost every industry; but, working at a time when open source frameworks like Databricks were the hottest thing to invest in, he assumed that his company would be open source first and gradually move into machine learning.
“But as we started going for funding early on, we realized that none of the companies were entering the enterprise outside of tech companies. They were selling to technology companies, but not really big corporations and retailers,” said Liu.
“The volume of data, amount of access to clients, and just the existing technical depth and talent we have in different groups is a very unique.”
“We eventually evolved from just a data framework company to a full stack data framework and machine learning company. At the time, we didn’t call it machine learning, we just called it predictions. We were going to make predictions with the data. I think it’s only in the last couple of years that machine learning has become a hot topic from a vocabulary perspective. All machine learning really is is being able to make a prediction without a human, and with a lot of data.”
But Liu adds that machine learning processes will have to be vertical-specific; the processes that one uses for predicting drug reactions don’t apply to finance or insurance. So the Toronto company is dead-set on tackling retail giants bogged down by legacy infrastructure, terabytes of data, and hundreds of billions of dollars in transactional data.
Part of Rubikloud’s philosophy is that all machine learning has to happen in an elastic environment, and live in the cloud. Eventually, all companies will migrate to infrastrucures like Amazon Web Services, Microsoft Azure. So the company offers products like Rubicore, which migrates massive volumes of legacy data from large retailers into a cloud infrastructure, giving retailers a single clean source of data.
“We work with very large retailers and it takes a while to build out that relationship. But when we build it out, it’s very strategic partnership. We spend a lot of time investing in them, and vice versa. So it’s hard to remove us at that point,” Liu said.
Its other two products are more focused on day-to-day business operations. Rubikloud’s lifecycle manager helps retailers optimize their loyalty programs by analyzing specific customer behaviours, and a merchandising manager that allows companies to predict how a mass promotions and seasonal product offerings will affect a company’s revenue.
“To say that Toronto is the best place in the world for tech talent is a true statement, but an oversimplification.”
“We’re giving you that extra comfort level to test your gut instinct, and in many cases quantify what’s going to happen before you take a big risk of ordering 10 pallets of a product you think will sell at a discounted price,” Liu said. “When we say machine learning, we don’t mean visualization, analytics, or reporting; we’re going deeper and making predictions. In some ways, the machine is actually creating a campaign, setting a price point, setting a profit projection, and the company then executes it.”
As the current 55-person team grows to 100 by the end of the year, Liu says his decision to build the company in the city is strategic rather than Toronto pride for the sake of it.
“To say that Toronto is the best place in the world for tech talent is a true statement, but an oversimplification,” said Liu, who adds that a lot of technical talent coming out of the University of Toronto and Waterloo in the past five years have gone to big tech companies like Google, or retailers and banks. “For us, this is a great place to recruit seasoned mid-career level engineering talent.”
As a non-technical founder, Liu being able to attract highly-specialized talent like data scientists, infrastructure engineers, and product managers is key for a startup like Rubikloud tackling machine learning. And while he lacks the deep tech expertise, he says that he makes it up in technical confidence and promising potential employees they’ll work at a startup that’s laser-focused on becoming a strong machine learning-based startup.
“My admission that I’m not a coder and I don’t try to pretend to be one, is one of the reasons that we hired really great people that are experts in their field. I’m very open when I talk to engineers and I say I’m not an engineer, but we have great engineers and I understand the tech and its impact. This obsession with trying to be an engineer when you’re not is not good for startups,” said Liu.
“The volume of data, amount of access to clients, and just the existing technical depth and talent we have in different groups is a very unique. It’s kind of a mini group within a larger organization. If IBM or Oracle or Google carved out a machine learning lab for retail, you would end up with DNA of what Rubikloud looks like.”
As for his long-term vision for the company, Liu said he believes that Rubikloud can get to $100 million in revenue in the next five years.
“The size of the contracts, the size of the retailers, and the board level support we get when we go in… we’re not mass SaaS. We’re not Hootsuite or Slack where we’re going to have hundreds of thousands of customers. For us to get to $100 million in revenue, we’re going to have maybe 75 clients,” he said. “Infrastructure-as-a-service providers like Oracle, Google, and Amazon are in a race to win the cloud and we think we have a huge head start on that when it comes to retail.”