MLOps ties in to Driverless AI and allows you to export your Driverless AI experiments into a MLOps project that can be viewed and shared from a single site. It also allows you to import, view, and share regression and classification models from H2O-3.
MLOps provides you with full model traceability: you can track a prediction for a specific model and investigate the report to understand how it was created.
A model can be deployed as a Staging, Dev, Production, or Challenger model. You can choose your best model (“Champion”) and continuously compare that model to a Challenger model. In addition to comparing models, MLOps also allows you to monitor models with checks that include drift detection and anomaly detection. And Model Endpoint links make it easy for you to retrieve the deployed model endpoint (with a provided sample cURL request).