Hackathons are like Olympic-level gymnastics competitions but for software developers. Competitors set aside their lives and work like crazy with the hopes of coming home the victor. For a team of five software engineers from data science product company Arundo Analytics, a recent Hackathon at GE’s annual Minds + Machines conference in San Francisco meant virtually nonstop coding over a 30-hour period fueled by Red Bulls and Diet Cokes. But in the end, the judges were impressed, and their app, called Power Scheduler, won the top prize.
Power Scheduler serves a simple but important function: It generates an automated schedule for power plant machines, balancing the supply between how much energy is available and how much electricity is needed. “A lot of factories have, say, a 20-year-old piece of equipment and it’s only 60 percent efficient, and it doesn’t need to be running all the time,” says team leader Jeff Jensen.
Maybe think of it as Nest for industry. Nest performs the same function with a home thermostat that learns when its users are home and creates an energy-saving schedule based on that information. Arundo’s app reads live data from a source like a solar grid and then measures that against web-connected devices or machines that need that power to run. Using machine-learning technology, it determines the best time to turn the machines on and power them down.
The app runs on Predix, GE’s operating system for industrial apps, which increasingly enables machines such as jet engines or locomotives to be monitored remotely or even “talk” to one another.
Jensen thinks that his team was picked out of more than 20 others in part because they built their app from scratch, which was no easy feat in itself.
The team put the basic building blocks of the app together on their laptops in a vast, white tent on the outskirts of the Minds + Machines conference on Pier 48 on San Francisco Bay. As the Tuesday night chill settled in and their fingers became too numb to type, the team piled into an Uber car and headed back to the townhouse they’d rented across town near the city’s scenic Presidio district.
At 8 p.m., they spread their equipment out in the main room, ordered some pizzas and worked flat out until 2:30 a.m. A short rest later and the team convened at 7 a.m. to finish off the app, handing it in just in time for the hackathon’s 1 p.m. deadline on Wednesday. The finished product runs on the Predix platform and uses Predix analytics to authenticate users and track live data from sensors on machines.
A panel of judges from Intel and GE, with backgrounds in everything from engineering to marketing evaluated the app and put Power Scheduler in the top three. Each finalist then pitched their app to an audience of 500 Minds + Machines attendees. When the moderator announced the team from Arundo had won, co-founder Stuart Morstead stood and cheered.
The win is even more impressive considering that until a week before the event, the team didn’t even know how to code on Predix. Jensen says that GE’s free online training boot camp was enough to help them pull the app together. Sometime in the next few months the team will travel to Paris — a trip worth $50,000 that makes up their “best in show” prize — where they’ll work with senior Predix developers at GE Digital’s European Foundry in Paris to finesse their app. Down the line, the team plans to launch Power Scheduler on the Predix catalog.
“We believe in the open eco-system that GE is building and look forward to deploying the app in both the Predix and Arundo environments,” says Morstead.
While Arundo won the Hackathon, there were two runners-up who each went home with $25,000. The first, Azuqua, built an app called FLO Force Once, which automatically senses when machines are malfunctioning and finds and dispatches the closest maintenance technician. The other runner-up, called Ship Twin, uses Predix’s data and mapping services to monitor cargo ships and detect potential piracy threats.