Vai al contenuto

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

Related terms