Dheeraj Bhandaru

 Designed and implemented customer propensity models (revisit, purchase) capturing the top ~15% of customers responsible for ~55–60% of revisits and improving purchase precision from 5% to 12% with 78% recall, optimizing sales outreach and conversion.  Evaluated and selected optimal algorithms, including linear, tree-based, and ensemble models, with a preference for CatBoost for its accuracy, stability, and interpretability.

 Engineered a warranty claims scoring pipeline using statistical analysis and machine learning, leading to a reduction in fraudulent claims and saving $1.6M for a leading automotive client.  Implemented anomaly detection & predictive models, achieving an F1 score of 0.75 for real-time and batch claim submissions.

 Enhanced operations efficiency by 25% for food industry clients by analyzing sensor data and filtering irrelevant training images.  Cleaned and transformed data using advanced Excel functions and Python libraries (Pandas, NumPy, Seaborn) to support field service monitoring.  Created interactive Tableau dashboards to visualize KPIs and communicate insights to stakeholders.