ML Model Factory

Registered ML tasks available for training — task type, status, and configuration for each supported prediction objective.

Recent ML training runs — task, dataset version, status, and metric summaries across all completed and in-progress runs.

Recent model evaluation runs — artifact, status, and metric summary for each completed or in-progress evaluation.

Model artifacts nominated for production deployment — approval status, task key, and review state for each candidate.

Model health summary — scalar performance and drift indicators. Nested payloads are not expanded; approval and validation states always come from backend fields.