← Back to projects
data science · November 2025
Customer Churn Predictor
Placeholder — an end-to-end gradient-boosted churn model deployed behind a lightweight API.
Python XGBoost FastAPI MLflow
outcome
AUC 0.89
Problem
Placeholder problem statement — describe the business context, the data, and the constraint set that made this project interesting.
Approach
Placeholder approach — feature engineering, model selection, validation strategy, deployment shape.
Results
Placeholder outcome — metric lift, stakeholder impact, what shipped.