MLOps
A set of practices to bring ML models to production and keep them reliable.
MLOps is DevOps applied to machine learning: model and data versioning, deployment, monitoring, and automated retraining. Tools: MLflow, Kubeflow, Weights & Biases, SageMaker.
Practical examples
- Model versioning with MLflow
- A/B testing models in production
- Monitoring model drift
- Scheduled retraining