Social Change, Powered by AI

Isobar has a deep tradition of Hackathons and, in more recent years, they have turned into formal, annual events across all of our offices. Over the last four years, projects have lived on and evolved into great assets for Isobar and its participants. We firmly believe in “Ideas Without Limits,” which means that we continually push the boundaries of technology and creativity as we strive to produce transformative digital experiences. Hackathons foster our culture of creative problem solving to positively impact and invigorate the company’s spirit of innovation and collaboration. Employees get to be a part of something cool and interesting, where they get to interact with passionate creatives, designers, engineers, developers and strategists. Everyone is welcome to participate in Isobar Hackathons and the best teams are often multidisciplinary. Considering the event starts on a Friday afternoon and ends Monday morning, the participants are often the most enthusiastic and passionate employees at Isobar: the ones who want to change the world with their curiosity, creativity and technical skills.

Hackers love coming up with new ideas and taking risks; they love to make and break things and are not afraid to fail. A common misconception about hackathons is that you need to  invent something new. Of course, it’s always nice when something entirely original comes out of a hackathon, but teams don’t have to invent anything. Sometimes the outcomes are equally impressive when new approaches to solve real world problems, using pre-existing tools and libraries, lead to a new take on an older idea.

Judging is conducted by more senior executives across all offices. The teams are scored on the work that is done across a variety of criteria that ranges from design, value to the marketplace and potential value to our clients. In addition to how well the concept fits the theme, storytelling is crucially important. The judges are given five questions that they use to rate each presentation.

Last year, one of the Hackathon team projects, Common Ground VR, was showcased at Games For Change and we continued to develop it as more brands took notice. You can read more about that project in in Forbes. The year before that, we tasked the team to “Tell A Story Without Words.” One of the top teams continued to develop their “Mooderator” project into a full fledged Arcade version!

This year, we had high demand to solve a problem that could impact the world around us in a meaningful way. As such, for 2018, our 4th annual Hackathon was themed  “Create Social Change through Machine Learning and Artificial Intelligence.” Leading up to this years Hackathons, Isobar held a series of Workshops on Machine Learning. Specifically, we focused on TensorFlow and how to approach ML from a design perspective. In addition, roughly a dozen engineers took Tensorflow and other machine learning tutorials in the weeks before the Hackathon.

Hackers were asked to deliver: a functioning prototype, a project webpage and a demo presentation. The prototype can be a patchwork of disparate components so long as it clearly demonstrates the concept and functions well enough that the technical challenges, as well as the user interface, is well understood.

The Detroit team took home the prize this year with their project,“Distraction-Free Driving Trainer.”  The team focused on a formidable problem. “Distracted driving is a leading cause of motor vehicle accidents. 9 people die per day because of preventable distracted driving. 60% of teen crashes involve driver distraction.”  To solve this problem they used TensorFlow with ML5.js and an HD Web camera to track drivers and to measure when they were distracted.

Tightly following in second place was the New York team who derived inspiration for their project from United Nation’s Sustainable Development Goals. Specifically, the team tackled the Good Health and Well-Being and Responsible Production and Consumption goals, both of which are highly impacted by food. The solution, a smart food companion app called NUTRIADE (working title), decreases the barrier to establishing eating habits that are healthy not only for the individual but also for the world around us. The working prototype integrates two layers of AI (Computer Vision and ML) with live data from USDA to provide a personalized nutrition score for products just by snapping a picture of the packaging. The team plans to enhance NUTRIADE by integrating environmental impact data about the product packaging as well as by enabling near-real-time multi-product score comparison.

In third place, inspired by the Yellow September (suicide prevention month), Porto Alegre’s office created “Lighthouse”. The project is an Omnichannel API-based AI capable of recognize suicidal speeches and suggest actions based on those. It is important to talk about suicide prevention and acknowledge that as an issue. 75% of people who attempted suicide did not look for help in the previous year. Lighthouse allows people to share content posted by their friends or children and give feedback about the song they are listening on Spotify, the Medium post they are reading or even their latest tweets.

The team also developed a progressive web app for psychologists train and improve the AI, making it more assertive every time. Wouldn’t it be amazing if you could help a friend or even your relatives by anticipating a possible suicidal tendency? Lighthouse is already running on demo mode and doing its best to help people.

The Chicago office took a different route and found a solution to an issue that plagues many of us in our day to day. “Is This Recyclable” was a light-hearted approach that helped to encourage and educate each other to recycle smarter using image recognition to determine if an object should be recycled or thrown away.

The Boston office created GloomGone to help people overcome seasonal affective disorder (SAD), a type of depression affecting 10-15 million Americans that is most pronounced during the short days and cloudy skies of the winter season.  GloomGone is AI/ML driven software that applies predictive analytics based on local climate combined with user inputs to offer users an empathetic “gloom forecast” that includes suggestions for what users can to do help themselves improve their mood.  To do this technically, the system generates an Amazon ML Regression model, requests some user data, adds the local atmospheric conditions, builds a “light profile” and fits that to the ML model.