Canadian Company Helps Others ‘Shoot for the Moon’

By: Yasmin Ranade

September 4, 2020

Crater Labs is a company that develops artificial intelligence (AI) and machine learning (ML) software that helps customers go beyond their current AI footprint. Interestingly, this Toronto-based research laboratory and consultancy firm positions its services as ‘putting profitable moonshots in the grasp of Canadian businesses.’

I had the opportunity to ask Khalid Eidoo, co-founder and CEO of Crater Labs, about his business, the technology it uses, and the kinds of ‘moonshot’ projects that most interest Crater Labs.

Khalid Eidoo, Co-Founder and CEO of Crater Labs

“Alexei Gavriline and I founded Crater Labs in 2017 on the basis that there wasn’t a consultancy out there that was looking at taking on moonshots or high Return on Investment projects for small -to -mid size enterprises (SMEs) in North America,” began Eidoo. “Instead, many machine learning (ML) companies were focused on small, simple analytics projects. We saw a need to use ML to explore the possibilities of deep learning to solve problems that scale for companies.”

Shared Eidoo, “With moonshots, companies can be very apprehensive due to the costs that go into taking on one of these high risk, high reward type projects; therefore, we designed a business model where we can take on these high risks to allow our clients to, in the long-term, differentiate themselves, while at the same time making it palatable from a cost perspective.”

Crater Labs collaborates with companies from various industries such as HR, finance, transportation, legal, and construction.

How does Crater Labs use ML and AI to help its customers?

“As a machine learning consultancy, we use artificial intelligence to provide systems the ability to automatically learn and improve from experience, without being explicitly programmed,” explained Eidoo. “We help bring our customers’ ideas to fruition by cost-effectively unlocking the power of their most valuable asset, their data, to drive innovation.”

Specifically, “Crater Labs relies on Pure FlashBlade for its storage infrastructure,” shared Eidoo. “Pure allows our researchers to analyze, process, and move data at will as they run their experiments and makes it easy to collaborate with clients on these large-scale projects. It provides us with unmatched speed, reliability, and scalability to deliver transformative results to clients faster.”

Moonshots With Impact

Crater Labs most enjoys exploring high ROI projects for its clients. “The moonshots that interest us the most are projects where we are able to unlock and capture the value of their data, to inform their decision making, create transformational products, and position themselves for growth through innovation, all through advanced machine learning.”

Added Eidoo, “By eliminating the cost and resource barriers associated with these ambitious projects we can help SMEs differentiate themselves in the market and give them a long-term competitive advantage, as opposed to just keeping up with the trends that exist today.”

I asked Eidoo to share with me an example of its moonshot successes, including one that detected a pandemic risk ahead of COVID-19.

‘Near the end of 2019,” began Eidoo, “we implemented a risk identification model for a client that detected an emerging pandemic risk in late November – months before COVID-19 took its toll on the country.”

“Capturing that data to identify risks has huge implications moving forward. COVID has shown the need for companies to have better risk models,” warned Eidoo. “We were able to address our client’s needs which included identifying risk triggers and supply chain effects. A similar model also allowed us to help healthcare providers and insurers recommend interventions by understanding the technical communication patterns of doctors and clinicians compared to how patients speak about their symptoms and conditions.”

“An additional moonshot we conducted,” remarked Eidoo, “was with an HR company who uses technology to facilitate and automate the interview process. Candidates go through an automated video interview where the recording is transcribed and anonymized and then evaluated by HR professionals to determine which candidates will be chosen for an interview.”

“An issue this created for our client was that existing speech-to-text solutions performed poorly when processing non-North American accented speech. This prevented them from being able to transcribe and automate applicants on the platform who speak fluent English albeit with an accent,” explained Eidoo.

“Using ML we employed a solution to stylize the voice of speakers with foreign accents, such that the accuracy of their accented voices would achieve the same level of recognition accuracy as North American accented speakers. This project saved the client tens of thousands of dollars per month in manual transcription costs.”


Leave a Reply

Your email address will not be published.