🩺CKD Ensemble Classifier ML-based Chronic Kidney Disease prediction using 24 ensemble models across 3 feature sets (All, RFE, Boruta). Built with FastAPI + React/TypeScript, featuring MICE imputation, Borderline-SMOTE, and 8 ensemble methods (RF, XGBoost, LightGBM, GBDT, AdaBoost, Bagging, Stacking, Voting). Replicates Rahman et al., 2024 methord.

Reproducible QRISK3 10-year cardiovascular risk computation pipeline using UK Biobank Instance 1 baseline data.

An end-to-end machine learning project for predict employee job change after data science training using LightGBM with advanced preprocessing (MICE, SMOTE), SHAP explainability, and an AI agent powered by FastAPI, Streamlit, LangChain, and Ollama to provide actionable retention insights.

Advanced predictive modeling workflow in R for diabetes risk assessment. Features Rubin's pooled GLM, LASSO feature selection, Youden's J statistic optimization, and thorough calibration diagnostics (Brier Score & Hosmer-Lemeshow test)