From Classroom Teammates to Hackathon Winners: MS-ADS Students Develop Responsible AI Prototype for Public Health Planning

Out of more than 6,000 participants and 800 qualifying teams representing 150 countries, University of Chicago MS in Applied Data Science students Morgan Klutzke and Binjie Wang earned first place in the Graduate Track (Master’s & PhD) of the 2026 U.S. Artificial Intelligence Institute (USAII) Global Hackathon.

The pair’s award-winning partnership began in the MS-ADS classroom. Klutzke and Wang first worked together in faculty instructor Francisco Azeredo’s Machine Learning course, completing homework assignments and the final project as teammates. Later, while taking Machine Learning Operations with Sanjay Boddhu together, Wang mentioned a hackathon he had discovered online, and the two decided to compete as a team.

For Klutzke, it was her first hackathon. Wang had previously participated in one, but the week-long global competition presented a new opportunity for both students.

The competition challenged teams to develop responsible AI solutions to real-world problems. Klutzke and Wang selected the “Cost of Doing Nothing Simulator” challenge, which asked participants to model the long-term social and economic consequences of delayed intervention and help decision-makers evaluate different policy scenarios. 

Building the Opioid Action Engine

The pair ultimately focused on the opioid crisis, an issue that resonated with Klutzke because of its effects in her home state of Indiana.

As they researched the topic, one question kept coming up: How could they help decision-makers see the bigger picture? 

In their research, they examined individual aspects of the opioid epidemic, from healthcare costs and financial impacts to effects on families and children, but the team never found a tool that brought those perspectives together into a single, holistic view.

“We thought we should combine all the research together and see if we can simulate the total cost of this issue,” said Wang.

The result was the Opioid Action Engine, an AI-powered decision-support prototype designed to help policymakers and community leaders explore the long-term societal costs of opioid misuse, compare policy scenarios, and better understand the potential impacts of intervention over time.

Rather than making recommendations on its own, the platform combined simulation modeling, AI-assisted decision-support tools, and legislative research to help users evaluate evidence before making decisions.

Designing for the End User

As the project evolved, Klutzke and Wang realized that technical sophistication alone would not make their submission stand out.

Instead, they focused on the people who would ultimately use it.

Throughout development, they envisioned a hypothetical county public health director trying to understand the impact of opioid misuse within their own community. That perspective shaped everything from the dashboard interface to the platform’s plain-language explanations and tooltips.

“We were talking about the end user, such as a hypothetical county health director, all the time throughout the project,” said Klutzke. “I feel like we did a really good job of making something that’s actually usable by a person who doesn’t need to be well versed in data science. We have tool tips to support them and explain things in simple terms without coming across as condescending.”

Wang believes that shift in thinking became one of the team’s greatest strengths.

“Initially we were thinking about the problem, but at some point we realized we should think about users, not the problems,” said Wang. “We redesigned parts of the solution to make it more user friendly, and I feel like that mindset made a difference.” 

Responsible AI at the Center

Because the platform was designed to support public policy discussions, Klutzke and Wang knew accuracy and transparency had to come before automation.

One of the biggest risks they identified was AI generating authoritative-sounding legislative language that wasn’t fully supported by the underlying data. To address that, they intentionally designed the platform so that AI could assist with drafting language, but only within guardrails established by the simulation results. Policymakers remained responsible for interpreting the data and making every final decision.

Rather than relying on opaque “black-box” models, the team chose a Monte Carlo simulation, which estimates how different policy decisions might play out by modeling many possible scenarios. The approach allowed users to understand how projections were generated while keeping the results transparent and grounded in evidence.

“It’s the people who are looking at the output of the simulation and thinking about their local communities and what possible actions they can take,” said Klutzke. “The AI is just there to assist them in their own thinking. It’s not there to think for them.”

Collaborating as a Team

With just two team members, collaboration was constant.

Wang developed the backend infrastructure, including the simulation engine, APIs, and database architecture, while Klutzke designed the dashboard and user interface. Throughout the week, they collaborated over Zoom, continually refining the project as new ideas emerged.

Midway through the competition, they paused to evaluate their progress.

“We checked the requirements from the hackathon and thought that what we had was too simple, and that we needed to incorporate more function in it,” said Wang.

Rather than settling for a project that simply modeled outcomes, the team expanded the platform into a more comprehensive decision-support system. They added an AI-assisted legislative workflow that allowed users to draft potential policy responses, evaluate their projected long-term impacts through the simulation engine, and compare different scenarios before making decisions.

Lessons From the Classroom

For both students, the competition reinforced lessons they had already begun learning in the MS-ADS program.

Klutzke credited the MS-ADS Leadership and Consulting class, which helped her think beyond technical implementation and focus on defining problems, collaborating with teammates, and understanding the needs of end users. “It’s actually so important to teach you how to define a problem, how to work with a client, and how to work on a team with other people,” she said. 

Wang believes today’s AI tools have lowered many of the technical barriers to hackathons, making creativity and problem framing more important than ever. “The technical barriers to joining a hackathon are getting lower and lower,” said Wang. “What sets people apart is their ideas.” 

Although she was initially hesitant to compete, Klutzke hopes other MS-ADS students will take the leap. “You absolutely can do it, and it will be a huge confidence boost,” she said. 

For Klutzke and Wang, what began as a classroom partnership ultimately became a winning project. The team received a $2,500 cash award and a USAII scholarship for the Opioid Action Engine.

Interested in the technical details? Read more about the Opioid Action Engine, including the team’s architecture, and watch a demo on their Devpost project page.