GFT Takes First and Second at the Cambridge Associates FedEx Cup Hackathon
Each explored cutting-edge AI applications, resulting in a first-place finish, second-place win and learnings and insights to share within GFT.
The Cambridge Associates FedEx Cup brings together developers from across the tech industry and challenges them to think outside of the box and beyond their day-to-day roles. For GFT, it’s an opportunity to explore creative problem solving, strengthen team culture and continue to shape the future of innovation.
First Place Winner: Guilherme Albani Camargo’s Cat Classifier
Representing GFT as a one-man team, Guilherme Albani Camargo—currently studying for his bachelor's degree in computer science in Brazil—brought home first place with a project as fun as it was technically impressive: A machine learning (ML) model trained to recognize housecats in video frames.
Using YOLOv11m for object detection and Roboflow for image classification and annotation environment, Guilherme built his classifier with an image classification dataset of over 300 cat pics sourced from Kaggle, along with footage of his girlfriend’s cat, the project’s muse. His resulting model is already adept at identifying housecats, but he seeks to improve it by training it on more diverse data. For example, things that are not cats, like cars or trees, and things that are cats, but not housecats, like lions and tigers. He also wants to train it on pictures of housecats with different color filters, to ensure that it understands what a housecat is regardless of the context.
Guilherme’s project reflects a thoughtful approach to ML, focusing on generalization, nuance and real-world data, and it highlights our support of team members at every level to experiment, learn and lead.
Second Place: The Crash Catcher
Our second-place team, featuring Thiago Castilho, Vinicius Campos, João Gustavo Kmiecik and Adriano Mulinari, and their project, the Crash Catcher, earned second place by streamlining one of the most time-consuming tasks for developers, debugging.
Crash Catcher is an AI debugging tool that automates the error investigation process in web applications. When a client-side error occurs, the user’s browser sends an alert to Crash Catcher, which then gathers logs, captures user interactions and queries Amazon Neptune to trace relevant classes and methods. From there, AI analyzes the data and delivers a structured report, often including suggested fixes.
The team envisions a future state which instead of combing through bug reports and error logs, a developer can review Crash Catcher, immediately understand what went wrong and know what to do to fix it. If Crash Catcher recommends a fix, it can run on its own and the developer can authorize it to handle the fix itself.
“The idea,” said Mulinari, “was that no one has to do anything, just approve what AI is trying to do.”
The team wants to further develop Crash Catcher to integrate with Jira and ServiceNow, offer a live error dashboard and auto-generate pull requests with potential solutions. What sets their project apart is that it isn’t a tool that replaces developers, rather, it frees them up to focus on building.
Beyond the product itself, the Crash Catcher team’s collaboration reflected well on GFT. With years of hackathon experience between them, they’ve made it a mission to encourage more colleagues to participate, even organizing a post-hackathon showcase back at the office to share insights and inspire others.
Innovation, Inside and Out
Events like the Cambridge Associates FedEx Cup go well beyond competition. For participants, they offer a space to experiment, learn new tools and tackle challenges from fresh angles. For GFT, they are a chance to reinforce our culture of innovation, celebrate employee curiosity and invest in continuous growth.
“It’s not only to learn new skills,” said Campos, “but to challenge ourselves to see how much further we can start a new endeavor and a new project.”
At GFT, innovation isn’t limited to what we deliver to clients, it’s part of how we grow as teams, individuals and individuals of teams. Whether it is AI code debugging or teaching machines to observe the world, our people are building what’s next.
