Production-ready MLOps project for customer churn prediction featuring end-to-end ML pipeline, Docker containerization, DVC-based data & model versioning, and automated CI/CD with GitHub Actions for reproducible training, testing, and deployment.

Containerised a production-ready churn prediction REST API using FastAPI + Docker, training an XGBoost classifier with 91% AUC; served via 3 endpoints (/predict, /retrain, /health) with Pydantic schema validation, confidence score outputs, and <100ms response time.

Machine learning pipeline to predict customer churn for a telecom company. Includes exploratory data analysis, feature engineering, class imbalance handling with SMOTE, and a Random Forest classifier. Served via a FastAPI REST API and containerized with Docker.