At SAP Edunet Foundation, I contributed to machine learning model development, including preprocessing, feature selection, and training and evaluation. Through iterative feature engineering and hyperparameter tuning, I helped improve model accuracy from 72% to 78% and assisted with cloud-based deployment.
I designed an end-to-end platform to predict customer churn and forecast Customer Lifetime Value for over 4,300 active digital storefront customers. I combined BG/NBD and Gamma-Gamma models with XGBoost, then built a Power BI dashboard to segment customers by risk; in my final-year project, I also built a multimodal emotion recognition system using facial and speech analysis.

