In a landscape where well-funded companies with substantial computational resources dominate, student founders must rapidly test their ideas to stay competitive. Access to computing power is crucial for processing large datasets and creating products and services that successfully leverage generative artificial intelligence.
At the University of Waterloo, student founders are rising to the challenge, harnessing Velocity’s graphics processing units server to turn their visions into reality.
Storing massive data sets and models, a GPU server runs complex algorithms and retain large datasets for processing calculations. With its cloud-based options, GPU’s are available and while purchasing and managing a server is an option , it requires secure network infrastructure and regular maintenance to operate effectively.
According to Faculty of Engineering undergraduate students Dhriti Gabani and Suhani Trivedi, GPU purchasing is cost-prohibitive and can hamper innovation.
Their startup Wave aim s’increas e the speed and accuracy of epilepsy and seizure diagnosis. This potentially revolutionary product requires training large sets of brain wave data test and build a minimum viable product (MVP). With this requirement, Gabani and Trivedi turned to Velocity’s GPU server, utilizing its resources to make this critical process possible.
"We are training a foundational model," Gabani says. "Training this model requires GPU access, which is currently in high demand, but is a core piece of the infrastructure needed to move forward and kickstart our idea." Employing Velocity’s servers, Gabani and Trivedi have overcome this obstacle and beg u n training without barriers.
With their startup technology, Wave aims to create a software product that will someday become integrated into hospital and clinic systems. Clinicians would be able to run electroencephalogram results through the Wave software, speeding up the diagnostic process and improving result accuracy.
After requesting access to publicly available EEG data, Trivedi highlights that with the sheer size of these data sets, processing times are large and require repetition, leading to escalating costs. With GPU access, the Wave team can face this major roadblock with confidence."
Seeing their ideas come to life with the MVP they developed during their Enterprise Co-op at the Conrad School of Entrepreneurship and Business, Wave is poised to bring their product to clinics and independent neurologists for real-world testing and iteration.
With promise on the horizon, Gabani and Trivedi were victorious in Velocity’s Cornerstone 10-day Validation sprint and The Conrad School of Entrepreneurship and Business’ E-launch pitch competition, securing $10,000 in funding for their work on Wave.
Wave isn’t the only company accelerating its progress thanks to Velocity’s GPU. Student founder and mechanical engineering undergraduate student, Humza Ahmed says beyond the massive cost savings, access to Velocity’s GPU server comes at a pivotal time for his startup.
Ahmed won the spring 2024 Velocity Pitch Competition AutomaxAI , a web application that gives property appraisers neutral, data-driven analysis about how much a property is worth, using language and terminology specific to the industry.
AutomaxAI grows its customer base, it currently has paying customers in Pennsylvania, Ontario and Alberta, they plan to use the GPU server to improve product offerings and include image processing to analyze photos that appraisers take during inspections.
"For us to be able to train computer vision models allows us to be more competitive, which we couldn’t do without a dedicated server," Ahmed says. "Having this access accelerates our growth, allowing us to operate more efficiently and cost - effectively, especially as our user base continues to grow.
"This access opens avenues to experiment and refine ideas which would otherwise be too costly to pursue," Ahmed says.
The GPU server is open to all students at the University of Waterloo working to develop ideas into a business.
It does compute: student founders build and iterate faster with Velocity’s GPU server
Advert