Resume

• Provide statistical consulting and data analysis support for WVU and WVU Medicine faculty research projects
• Collaborate with clinical researchers on study design, statistical analysis, and manuscript preparation
• Methods include propensity score matching, survival analysis, Bayesian modeling (Stan), and causal identification (DAG design, regression discontinuity)

• Built full-stack application for a nonprofit in the animal welfare space
• Built full-stack application for a nonprofit consultancy

• Built LLM-based data processing pipeline for structured information extraction from legal documents
• Applied supervised learning techniques to predict outcomes of circuit court decisions
• Built web app for visualization and inspection of results

• Led development on full-stack application prototype in the education space using Next.js
• Built RAG system for generation of educational content and built front-end and back-end systems
• Designed and implemented hybrid LLM/traditional recommendation system using few-shot learning and collaborative filtering

• Built intuitive web-based interface for human-in-the-loop machine learning system
• Collaborated with data science team on design of LLM-based recommendation pipeline
• Prepared demo for strategic partnership pitch to Firefox, the only intern selected to do so

• Headed three-person front-end development team, developing project prototype in three months
• Faced with performance issues, developed custom data structure that cut render times by 90%
• Iterated on product-market fit by conducting 5+ user-interviews, leading to a pivot
• Presented to CEO of Deepcore and VP of SoftBank, resulting in continued funding for the project

Vettiyil B, Wen S, Ware L, Ashikyan O. Glenohumeral septic arthritis in intravenous drug users: 10-year retrospective analysis. J Clin Imaging Sci. 2026;16:18.

Watson et al. Clinical Outcomes of Aspirin Plus Clopidogrel Versus Aspirin Plus Ticagrelor in Left Main Coronary Artery Stenting: A Retrospective Cohort Study

Lanehart et al. Lipomas Treated with Subcutaneous Injections of Cooled 0.9% Sodium Chloride: A Nonrandomized Clinical Trial

Random Forest Models on Repeated Measures Data

• Collaborated with researchers in the WVU agricultural sciences department to predict gold-standard cattle weights from a novel, lower-cost measurement method using repeated measures data
• Developed novel random forest training system for time series and repeated measures data; designed custom validation pipeline showing it decreased overfitting and improved performance

Undergraduate Capstone: Polynomial-Time Causal Identification

• Found bug in Microsoft Research's DoWhy library (~50% of significance tests invalid)
• Implemented fix; developed version 60% faster than Microsoft's implementation

• 3.9 GPA

• Concentration in Data Science and Statistics