mlops
FinRisk AI - Production MLOps Credit Risk System
Built a complete end-to-end MLOps pipeline for credit default prediction with real-time drift monitoring and sub-millisecond inference.
June 1, 20266 technologies
About the Project
Built a complete end-to-end MLOps pipeline for credit default prediction, incorporating experiment tracking (MLflow), hyperparameter tuning (Optuna), model registry, CI/CD to Azure, and real-time drift monitoring. Delivered production-ready, explainable AI with SHAP transparency and sub-millisecond inference via FastAPI.
Key Features
- ✓Responsive design that works on all devices
- ✓Optimized performance with lazy loading
- ✓Accessible and follows WCAG guidelines
- ✓Clean, maintainable codebase
Technologies Used
FastAPILightGBMMLflowDockerAzureEvidently AI