Conducting academic research at the Computational Thinking Lab, an AI research group at Indiana University's Luddy School of Informatics, Computing, and Engineering. The lab's mission is to make healthcare safer, more private, and more effective, organized around three pillars: Smart Medicine (predictive systems that catch disease biomarkers before clinical symptoms emerge), Human-Centered AI (personalized, behaviorally-grounded wellness models), and Green AI (low-compute, energy-efficient healthcare systems).

My work sits primarily in the Smart Medicine track. Recently I have been fine-tuning open-source large language models — both general-purpose and models pretrained on medical corpora — for tasks such as predicting diagnoses from patient data and extracting structured clinical information from unstructured electronic health records (EHR).

A core part of the work is evaluation: comparing base models against fine-tuned variants across different extraction and prediction tasks to understand where fine-tuning actually moves the needle versus where prompting alone is sufficient, in line with the lab's broader emphasis on rigorous, multi-dimensional evaluation of healthcare AI rather than single-metric benchmarking.

This work runs on IU's LAIR multi-GPU cluster, where I manage large-scale training and inference workloads — batching experiments across models and datasets, and optimizing for the compute constraints that come with iterating on multiple LLM variants at research scale.

In short, pretty exciting stuff!