Yashraj Jadhav

• Built a patient readmission classifier comparing Random Forest vs XGBoost on 500K+ records in scikit-learn; handled ~15% positive-class imbalance using stratified k-fold cross-validation and selected the final model by optimizing Precision-Recall AUC over default accuracy, cutting readmission rates by 18% across three hospital departments • Designed a causal inference pipeline for pharmaceutical trial evaluation: defined treatment/control cohorts, ran stratified subgroup analysis across demographic and clinical segments, and computed effect sizes with confidence intervals to support $4.2M go/no-go investment decisions • Applied K-Means clustering (elbow method + silhouette scoring) and Apriori association rule mining on 2.5TB of diagnostic records using Python and SQL; surfaced high-confidence co-occurrence patterns that reduced misdiagnosis rates by 23% • Engineered SQL-based feature extraction and ETL pipelines processing 50K+ daily records with schema validation, null-rate monitoring, and incremental loads; improved pipeline throughput by 35% and downstream feature quality by 28% • Delivered model predictions (risk scores, cluster assignments, subgroup effects) via Power BI decision dashboards to clinical executives; tied every visual to an upstream model output, reducing operational costs by 15%

• Profiled 300K+ patient records using Python and SQL to assess feature distributions, cardinality, null rates, and class balance; findings directly shaped feature selection and encoding decisions for the readmission and clustering models built in the subsequent full-time role • Conducted hypothesis testing and statistical analysis on patient safety event logs across 3 clinical units, identifying significant outcome patterns that improved clinical protocol adherence by 22%

Relevant Coursework: Machine Learning, Large Language Models, Advanced SQL, Applied Modern Statistical Learning Methods