Real time observability for ML models

Seldon Core provides out of the box Prometheus metrics for monitoring. These include operational metrics like request latency and feedback on predictions.

The Seldon Deploy UI provides a number of dashboards exposing both these metrics and more about model and deployment usage. You can also access these directly via the Prometheus UI or other tools that integrate with Prometheus, such as Grafana.

Detailed demos showing monitoring are available in the Demos.