AI breakthroughs and bold ideas shine at MSOE’s largest-ever Rosie Supercomputer Super Challenge
Congratulations to the winners of the 2026 Rosie Supercomputer Super Challenge! This year marked the fifth anniversary of the competition and it was the largest to date with 27 entries—nearly doubling the number of entries received last year. The winners were chosen after the six finalists presented their projects in front of a team of judges that included:
- Dr. Dwight Diercks ’90, NVIDIA senior VP of software engineering, MSOE Regent
For the annual challenge, sponsored by Diercks, MSOE students demonstrate how they’ve used Rosie the supercomputer to solve a problem, improve a process or answer a difficult question during the Rosie Supercomputer Super Challenge.
2026 Rosie Supercomputer Super Challenge Winners
1st Place: Adam Haile
2nd Place: Dylan Norquist
3rd Place: Josh Myers, Xander Ede, Eddy Chukwuma
First Place:
SkyNet: Belief-Aware Planning in Partially Observable Stochastic Games
Adam Haile, computer science and machine learning
This project uses reinforcement learning to demonstrate how an AI agent can be trained to play a game called SkyJo that incorporates hidden cards and random chance.
Second Place:
SMEARGLE: Sketch the Draft, Skip the Attention
Dylan Norquist, computer science and machine learning
This project introduces a more efficient mechanism for the transformer architecture (the AI model structure that underpins most major generative AI models).
Third Place:
Proactive Urban Forestry Management: A Machine Learning Approach to Predicting and Prioritizing Tree Pruning in Milwaukee
Josh Myers, computer science and machine learning; Xander Ede, computer science and machine learning; Eddy Chukwuma, computer science; Dylan Norquist, computer science and machine learning
This project explores how satellite data can be combined with forestry data from the City of Milwaukee to help re-prioritize how the forestry department manages the hundreds of thousands of trees across the city in a proactive and more cost-efficient manner.
Honorable Mentions:
From Revit to Robot: BIM-Driven Simulation for Autonomous Building Operations
Owen Pacetti, computer engineering; Steven Thomas, biomedical engineering; Joseph Loduca, software engineering; Nicolas Picha, software engineering; Tanner Cellio, computer science and machine learning; Diego Gonzalo, computer science; Delsoro Some, computer science; Adrian Manchado, computer science and machine learning
This project explores the use of a digital twin that was constructed for the new engineering building at MSOE from the blueprints. The digital twin was used to train a robot to help it navigate and understand the nuances of the new building prior to the building existing.
Teaching Agents how to Bargain at Settlers of Catan
Mazen Hamid, computer science and machine learning
This project explores the use of reinforcement learning to train an AI agent to play Settlers of Catan including the bargaining components of the game.
TerraCare: GPU Accelerated healthcare
Alhagie Boye computer science and machine learning; Wilfred Tapsoba, computer engineering
This project explores how satellite and other data can be used to intelligently identify medical deserts and place medical facilities strategically across Africa to improve accessibility and patient care.