Applying machine learning and data science to solve real clinical and industry problems — from predicting transplant outcomes to deploying models at scale.
I'm a Data Scientist specializing in applying machine learning to high-stakes domains — particularly clinical medicine, where predictive accuracy has real consequences for patients.
My research and applied work spans image processing, data extraction pipelines, and predictive modeling. I'm published in Kidney International Reports (Elsevier), where my work focused on predicting post-transplant outcomes using ML — work that directly informs organ allocation decisions.
I'm actively seeking opportunities to collaborate with research teams, industry partners, and organizations building data-driven solutions at scale.
Led comprehensive analysis of kidney transplant Medicare claims using the U.S. Renal Data System (USRDS), spanning 2000–2020 across 100,000+ patients. Identified cost reduction strategies with potential to save the government millions in healthcare spending.
Managed and analyzed over 1 billion rows of pre- and post-transplant hospitalization and physician supply claims in a HIPAA-compliant cloud environment. Insights are being used to understand drug consumption impact, optimize donor-recipient matching, and improve patient quality of life.
Extracted and analyzed data from PDF reports containing both computer-generated text and handwritten clinical notes. Data covered pre-transplant and post organ-failure tests, including blood pump metrics, heart rate, and oxygen monitoring at each timestamp.
Built a preemptive scorecard for credit card non-delinquent customers to optimize collection costs and restrict flow into higher delinquency buckets. Used bureau and demographic variables with roll rate and vintage analysis to define target variables.
Deployed logistic regression in SAS to identify potential customers likely to buy a product. Validated accuracy metrics, performed stress testing, and monitored model stability over time. Targeted top-3 deciles via scorecard-driven campaigns.
Identified cross-selling opportunities among existing customers through customer profiling and clustering based on product usage patterns. Surfaced actionable segments for targeted outreach.
Analyzed digital payment gateway customers across POS, web-based payments, and third-party apps, benchmarked against mobile and net banking users. Quantified digital channel value to inform investment decisions.
Built a linear regression model for a major gaming client to pinpoint causes of revenue decline. Developed A/B testing strategies for Android and iOS users to validate and deploy interventions at scale.
Leveraged twitteR in R and SQL on Redshift to generate 360° client-centric metrics — interest segments, marketing buckets, sentiment scores, and peak active hours — for a major e-commerce client.