Master’s in Applied Data Science Autumn 2025 Capstone Winners

The Autumn 2025 Master’s in Applied Data Science (MS-ADS) Capstone Showcase marked the culmination of students’ applied learning experience, where academic coursework converges with real-world problems and key research areas. Student teams led all aspects of their project development, applying advanced data science methods to challenges spanning healthcare, sports analytics, mobility, enterprise AI, and global workforce strategy.

Across sessions, judges recognized a select group of projects as Best in Show, highlighting work that demonstrated strong technical execution, thoughtful problem framing, and the ability to translate analytics into practical, usable solutions 

Alumni Judge Perspective

Prashant Kulkarni (MS-ADS ’25) returned to the program as an alumnus and served as a judge for the Autumn 2025 Capstone Showcase.“It was great to be back with ADS as a capstone judge. A lot has happened for me since I graduated in the summer, and now. I started teaching at UCLAx and conducting research in AI Safety. My view has certainly progressed, and so has our industry. 

Kulkarni observed that the student projects that rose to the top were those that demonstrated a full research-to-production pipeline. Students showed they could move beyond theory to working code, execute with technical rigor, and communicate their findings clearly. In today’s fast-paced AI market, bridging the gap between theory and working code is a critical skill—and the students demonstrated they could do just that.

Best in Show Projects

Conversational AI | PRISM – Personal Response Interview & Situation Model

Presenters: Karim Derbali, Muhammad Hassaan Sohail, Sourendu Saha, Ankit Agrawal
Faculty Advisor: Nick Kadochnikov

This project focused on building a conversational AI system designed to model interviews and situational responses, reflecting the growing role of natural language interfaces and intelligent agents in real-world applications.

Chicago White Sox | Measuring Pitcher Deception and Command Using Motion Tracking Data

Presenters: Kurt Fischer, Manas Vemuri, Benjamin Brown, Vincent Chirio
Faculty Advisor: Jeanette Shutay

Partnering with a professional sports organization, this project applied data science techniques to motion tracking data to evaluate pitcher performance, illustrating how analytics can inform decision-making in high-performance sports environments.

Argonne National Laboratory | Dual-Agent Automation of Data Analysis for EVREZ

Presenters: Ritai Na, Anusha Bhat, Nidhi Pareddy, Xiao Wu
Faculty Advisor: Igor Yakushin

This project explored automated data analysis through a dual-agent framework, demonstrating how intelligent systems can support scientific workflows and accelerate analysis in research-driven settings.

Argonne National Laboratory | Extending ARGO with a Knowledge Graph

Presenters: Andrew James, Samuel Fisher, Samuel Park, Eric Huang
Faculty Advisor: Igor Yakushin

Focusing on knowledge representation and structured data integration, this project extended an existing platform with a knowledge graph to improve data connectivity and analytical insight within a national laboratory context.

GPT for Healthcare | Applications of Generative AI in Emergency Department Admission Evaluation

Presenters: Cassandra Chen, Monica Ko, Jane Lee, Alvin Yao
Faculty Advisor: Utku Pamuksuz

This healthcare-focused project examined how generative AI can be applied to emergency department admission evaluation, highlighting the role of advanced AI systems in supporting complex, time-sensitive clinical decision-making.

Global IQ | Optimizing ROI of Global Talent Mobility Through Predictive Modeling and Optimization for Corporate Programs & Policies

Presenters: Lin Zheng, Alex Ding, Zhiqi Gao, Lejun Liu
Faculty Advisor: Gizem Agar

Addressing challenges in global workforce strategy, this project applied predictive modeling and optimization techniques to help organizations evaluate and improve the return on investment of talent mobility programs.

HERE Technologies | Ensuring Map Freshness with an AI-Powered Places Data Pipeline

Presenters: Akanksha Mathpati, Soham Mandal, Aarav Vishesh Dewangan, Junfan Zhou
Faculty Advisor: Sanjay Boddhu

This project focused on building an AI-powered data pipeline to maintain accurate and up-to-date place data, demonstrating how automated analytics systems support large-scale geospatial platforms.

Inference Analytics | Automated ICD-10 Code Prediction for Healthcare Reports Using Large Language Models

Presenters: Apoorva Prakash, Swayam Desai, Zheyi Liu, Jiashu Yuan
Faculty Advisor: Utku Pamuksuz

Applying large language models to healthcare data, this project addressed the automation of ICD-10 code prediction, reflecting industry demand for scalable solutions that improve efficiency and consistency in clinical documentation. 

One of the Best in Show projects, GPT for Healthcare, offered a closer look at how these skills come together in a real-world, high-stakes setting.

Best in Show Project Spotlight

GPT for Healthcare | Applications of Generative AI in Emergency Department Admission Evaluation

Presenters: Cassandra Chen, Monica Ko, Jane Lee, Alvin Yao
Faculty Advisor: Utku Pamuksuz

The GPT for Healthcare team explored how generative AI can support emergency department admission evaluation, a setting where rapid, high-stakes decision-making is critical. Their work focused on agentic AI workflows designed to integrate both structured medical records and unstructured triage notes while remaining adaptable to different hospital environments.

The team emphasized that agentic workflows offer a unique blend of modularity, customizability, and iterative improvement, allowing the system to integrate with existing machine learning models and knowledge bases already in use across hospital networks. Because different hospitals may have existing ML implementations or access to different databases, they noted that these systems can serve as customizable touchpoints for the agent to access. The team also added that human feedback checkpoints can be tailored to each emergency department’s preferred risk thresholds.

Winning Best in Show, the team said, reflected more than a final result. “This award is an affirmation and recognition of the hard work everyone on the team has put in over the past twenty weeks.” The team faced challenges, but with guidance from faculty advisor Dr. Utku Pamuksuz and teaching assistants Tegan Keigher, Andrew Alvarez, and Mary Erikson, the team was able to re-triangulate which intersections they were able to investigate and push medical technology forward.

Taken together, the Best in Show projects offered a snapshot of how applied data science is practiced today, not as isolated modeling exercises, but as work that must operate within real constraints, messy data, and real decision-making environments. The Autumn 2025 Capstone Showcase made clear that these students are already grappling with the kinds of challenges they will face beyond the classroom.