Master’s in Applied Data Science Summer 2025 Capstone Showcase
At the Master’s in Applied Data Science (MS-ADS) program’s Summer 2025 Capstone Showcase, eighteen teams presented solutions that brought together classroom learning, advanced research methods, and real-world problem solving. This showcase reflected a cross section of project types within the program, including research-focused, industry-generated, and student-proposed work across industries from healthcare to finance to energy and agriculture.
The judging panel, composed of industry leaders and alumni, was impressed by the quality of work, noting that many teams applied leading-edge analytic solutions, often leveraging tools and technologies beyond the standard coursework and, when faced with data limitations, developed inventive, practical solutions that added measurable value for clients. One of the judges, Devdutt Sharma (MS-ADS ’23), shared: “What stood out most was the professionalism of the presentations and how thoughtfully students connected their technical work to real business problems. The teams didn’t just build models; they thought about integration into client workflows, measurement of success, and the path to production.”
The opportunity proved to be a transformative experience for students. As one member of the Best in Show-winning PowerKiosk team reflected: “We learned that one of the hardest parts of building AI solutions is not just model accuracy, but making them fit into real workflows. Balancing technical innovation with practical adoption was key to delivering lasting impact.”
Learn more about the projects that won Best in Show below.
BEST IN SHOW WINNERS
PowerKiosk | Automated Utility Bill and Contract Processing Using Agentic RAG
Presenters: David Zhu, Joseph DiGiovanni, Amy Tramontozzi, Fletcher Lin
Faculty Advisor: Utku Pamuksuz
Project Topic: Partnering with PowerKiosk, an energy procurement start-up, this team addressed the labor-intensive and error-prone task of manually extracting and analyzing utility bills and contracts. Their AI-driven solution combines Optical Character Recognition, Computer Vision, and Retrieval-Augmented Generation to interpret customer information, tabular and graphic usage data, and compute rates required for financial analysis. Compared to the company’s internal prototype, the system achieved higher accuracy and faster processing times, seamlessly integrating into PowerKiosk’s platform to deliver more timely and reliable business reviews and budget forecasts.
UChicago Medicine | Physics-Informed Deep Learning for Prostate Cancer Detection
Presenters: Ethan McManus, Zoe Toy, Leon Tan, Eric Dick
Faculty Advisor: Batu Gundogdu
Project Topic: This team collaborated with UChicago Medicine to develop a machine learning model that uses physical rules to improve prostate cancer detection from MRI scans. By incorporating physical rules into a machine learning framework, the model is able to estimate key tissue properties with greater accuracy without a biopsy, even in low-quality imaging conditions. Compared to standard nonlinear least squares methods, their physics-informed autoencoder delivered superior results, highlighting the potential of physics-informed deep learning to improve medical imaging and reduce the need for invasive procedures.
Grass Fed Valley | Meat Market Forecasting & Pricing
Presenters: Natalie Foemmel, Brandt Buchda, Yijia Song, Arpan Pradhan
Faculty Advisor: Dmitri Sidorov
Project Topic: Grass-fed beef producers often lack access to real-time pricing and forecasting tools, relying instead on fragmented data sources such as USDA reports, ecommerce listings, and internal sales logs. Partnering with Grass Fed Valley, this team built a weekly-updating data pipeline that scrapes, standardizes, and aggregates pricing data across 56 animal products. By applying time-series forecasting and retrieval-augmented generation, their system provides clear visibility into pricing trends and supports smarter price planning. The project demonstrates how lightweight, automated intelligence can help close the analytics gap between small producers and industrial-scale operations.
A Showcase of Applied Data Science
Three times each year, our Capstone Showcase demonstrates the transformative potential of applied data science. The Summer 2025 Capstone Showcase continues that tradition. From healthcare diagnostics to energy automation to agricultural forecasting, the winning projects and the strong showing from all teams reflected the program’s mission to prepare graduates to innovate, collaborate, and drive meaningful impact across industries.