Every hackathon is literally a race against time, regardless of where they’re held, the skill level of the participants or the technology involved.
When Daisy recently hosted a hackathon with the University of Toronto’s Division of Engineering Science, however, we presented students with a particularly challenging problem that put their mental pedals to the metal.
Over a 24-hour period, we asked the students to apply artificial intelligence to determine the best way to get a racecar around a track. The winners would receive a paid internship, cash prizes, and bragging rights.
It did not involve a real race car, of course; we provided the 60 students from engineering science, engineering and computer science that entered CSV files with a series of points that showed every curve of eight simulated F1 race tracks, along with instructions to show how they could configure their race car.
As anyone who has watched F1 racing knows, this is a lot harder than it sounds.
Much like the many factors that can affect performance in businesses like grocery and insurance, racecar drivers need to keep multiple things in mind without slowing down.
Among other things, it included determining the optimal car configuration (power, acceleration, grip, tire wear, fuel capacity) and the optimal driving instructions (where on the track and how much to accelerate, brake and turn)
Since it would take eons to work through every possible scenario, the hackathon offered a great opportunity for students to use optimization algorithms and artificial intelligence to deliver the best lap times across the 8 F1 tracks.
Where Math and Science Meets Creativity
Winning our hackathon wasn’t just about how fast the teams worked, but also how well they could show and discuss the results. Teams had to present short slide decks that detailed their methodology; along with any source code they had created during the hackathon. This scoring approach is important because even if you’re technically smart, success in business means communicating the value effectively to the rest the world.
The teams in this year’s event did a great job, but it’s worth spending a minute to talk about why this type of hackathon is important.
Building a better racecar might not seem comparable to the hurdles that retailers and insurance firms are trying to overcome every day.
And yet the challenge we presented the teams involved the same kind of thinking and problem solving that led to reinforcement learning in the artificial intelligence (AI) applications that are transforming businesses today.
Regardless of who wins in a hackathon, anyone who participates begins to see first-hand the kind of amazing careers that they can have in sectors from financial services to the grocery sector and beyond.
Daisy’s CEO, Gary Saarenvirta, believes that identifying those sorts of opportunities is critical to the next generation of knowledge workers.
“I walked into this career accidentally,” he admitted in a conversation following the hackathon. “My hope is that by hosting this hackathon for students, it doesn’t have to be a happy accident for them. They can choose this kind of work more consciously.”
In other words, we want to help the future of A.I. talent start their engines because once they hit the workforce, there’s going to be no stopping them.
The power of parallel computing
After the hackathon, Daisy’s software engineering team used its computing infrastructure powered by NVIDIA GPUs and our reinforcement learning approach to deliver lap times that beat the best student results.
It was an unfair advantage for Daisy’s team but it showed that reinforcement learning powered by massive computing horsepower could beat the best human brains at tasks that are beyond human capability.
It was an exciting and successful event! We look forward to next year.