Designed credit scoring model for 500+ farmers using XGBoost and SHAP, achieving 89% predictive accuracy. Engineered in-app data pipeline aggregating agricultural and financial records across 6+ metrics (yield, irrigation, profit, etc.). Generated and validated synthetic datasets simulating farm scenarios to enhance model training and performance. Collaborated with developers and analysts to align scoring logic with fintech and agri-credit standards. Integrated weather and insurance variables into risk assessments, improving model fairness and adaptability.

Built Python-based financial models improving revenue forecasting accuracy by 22%. Automated reports for 15+ SMEs using SQL/Python, reducing processing time by 35%.