Job Management

The training job manager handles the full lifecycle of training jobs including scheduling, resource allocation, monitoring, and failure recovery.

Job Lifecycle

State	Description
`pending`	Job created, awaiting resources
`queued`	Job queued for execution
`preparing`	Loading data and initializing model
`running`	Actively training
`completed`	Training finished successfully
`failed`	Training failed with error
`cancelled`	Job was manually cancelled

GET /api/v1/training/jobs/{job_id}

Response includes real-time metrics, resource utilization, and estimated completion time.

File	Path
JobManager	`data-plane/ml-service/src/training/job_manager.py`
Job Scheduler	`data-plane/ml-service/src/training/job_scheduler.py`
Job Monitoring Service	`data-plane/ml-service/src/training/job_monitoring_service.py`