At Himmel Infotel Pvt. Ltd., I built regression models for house price and addiction risk prediction, achieving 80–90% accuracy, and developed a K-Means customer segmentation system with OpenCV image processing.
For my internship capstone, I built and delivered ChurnGuard, a cost-sensitive customer churn prediction system. It uses threshold optimization to minimize business cost rather than raw accuracy.
I developed ChurnGuard’s end-to-end MLOps pipeline with FastAPI, React, MLflow, SHAP, and Docker, including consistent preprocessing and explainability for interpretable predictions.
On a road accident severity prediction project, I compared several models and achieved 81% accuracy with Random Forest. I also built management systems during my Python internship at Geeta Technical Hub.
