MATIH Platform is in active MVP development. Documentation reflects current implementation status.
15. Workbench Architecture
Model Deployment

Model Deployment

The deployment components at frontend/ml-workbench/src/components/Deployment/ manage model serving endpoints, health monitoring, and rollback capabilities.


Components

ComponentFilePurpose
DeploymentDashboardDeploymentDashboard.tsxOverview of all active deployments
EndpointHealthCardEndpointHealthCard.tsxIndividual endpoint health metrics
RollbackModalRollbackModal.tsxModel version rollback confirmation

DeploymentDashboard

Displays all active model serving endpoints with status, latency, throughput, and error rate metrics.

EndpointHealthCard

Shows real-time health for a specific endpoint:

  • Request latency (p50, p95, p99)
  • Throughput (requests/second)
  • Error rate
  • Model version and stage
  • Resource utilization (CPU, memory, GPU)

RollbackModal

Confirmation dialog for rolling back to a previous model version with impact analysis and traffic migration options.

DeploymentManager

File: frontend/ml-workbench/src/components/DeploymentManager/index.tsx

Full deployment management interface with:

  • Blue/green deployment configuration
  • Canary deployment with traffic splitting
  • A/B testing setup
  • Auto-scaling configuration
  • Resource allocation