Matt's Formula E Hackathon Journey

I had the incredible opportunity to be invited by Google Cloud to participate in the world’s largest AI hackathon, held at the London Formula E event at the ExCeL Centre. It was an exciting chance for The Dot Collective to collaborate closely with our client, Places for People, as we teamed up to hack using Google’s latest AI tools.

As I walked through a futuristic light tunnel at the entrance, the scale of the world-record-breaking hackathon suddenly became clear. Just in front of the entrance, a sleek Formula E race car made a striking focal point. Beyond it, hundreds of tables stretched out, each occupied by hackers eagerly opening their laptops in anticipation.

The day began with a talk from Eric Ernst, VP of Technology at Formula E, who showcased the amount of innovation happening in the sport. Only in recent years, he explained, had the car batteries been able to last a whole race, previously drivers would jump from one car to another fully charged car mid race!

We then received a brief of the hackathon topics before it was ready, set, go! Well almost… turns out that 1,130 people all clambering on the WIFI at the same time made for a slightly slow start, but we soon got up to full speed.

Faced with a variety of problems to solve, ranging from real time crash detection to personalised highlights, our team chose to target fan engagement. We worked on an AI solution to chat with new fans and introduce them to a driver they'd be most suited to support. For example, suppose a Dutch Formula One fan has recently become interested in Formula E, our application might recommend Nyck de Vries, who has competed in both sports, and allow them to learn more about De Vries in a conversational manor.

In a few hours, we were able to create a working prototype which we live demoed to the Google judges. Our solution used Gemini 1.5 Flash to convert text to SQL, run queries, and incorporate the results into its chat responses. We're all familiar with chat applications hallucinating and confidently returning wrong answers, which is why this approach of combining the precision of SQL with the usability of a natural language interface is so compelling.

After presenting our solution to the judges, we headed over to watch the Formula E warm-up laps. Once our necks were worn out from watching the passing cars, we went on a pit lane walk, where we had the exciting opportunity to see the teams up close as they meticulously prepared their cars for the race day.

It was great to hack with the Places for People engineers for the day. As we work with them to build out the company's data capabilities on GCP it was exciting to talk about applying these solutions to their business, with wide ranging opportunities from improving call centre efficiency to improving searchability of internal documentation.

As an engineer focussing on building platforms, getting hands on with state-of-the-art models provides an invaluable view into the future of data pipelines which will serve AI use cases, for example, having seen the new advances in Gemini’s context window, it’s clear that video processing will soon become much more significant. I feel fortunate to work for The Dot Collective, a company who is heavily invested in staying ahead of the curve in a rapidly evolving industry.