Mynd for MLOps engineers
MLOps Engineer, as it will appear in the app, in the next version.
An MLOps engineer keeps models in production: deployments, monitoring and costs. The job is alerts, versions and incidents.
The ready-made structure
Score 87 out of 100. What the score means.
The structure starts with Models, the centre of the work, divided into Lifecycle, Versioning and Evaluation. Then come Pipelines, Infrastructure, Operations, Data, Governance and Costs. Each record keeps its full path, for example Models, Lifecycle, Development, and can be found by any of the three levels.
The full tree is at the end of this page, with the names as they are in the app.
Suggested Context phrases
The structure says what each record is about. The Context phrases, which each person creates, say the rest. For this profession, some that tend to serve:
Incident Deployment Cost To watch
A report that sends itself
A search gathers the month's incidents and deployments. On the first day of the month, the technical lead receives the summary.
What you gain: The production history lives in one place, ready for audit.
The card with the recovery procedure goes to the on-call team at the start of each rotation.
In this version, Start, each person has 14 automatic search sends and 14 card sends; in Run, they will be able to have as many as they need. Each send has its own search, its own recipients and its own frequency: daily, weekly, monthly or on a date. Emails go out from the person's own email account, and each recipient receives their own.
The full tree
Models3 Types
- Models/LifecycleDevelopmentDeploymentRetirement
- Models/VersioningModelsDatasetsExperiments
- Models/EvaluationAccuracyValidationMonitoring
Pipelines3 Types
- Pipelines/TrainingAutomationSchedulingOptimization
- Pipelines/DeploymentCICDRelease
- Pipelines/OrchestrationWorkflowsDependenciesCoordination
Infrastructure3 Types
- Infrastructure/ComputeCPUGPUScaling
- Infrastructure/StorageDatasetsArtifactsBackups
- Infrastructure/PlatformCloudContainersClusters
Operations3 Types
- Operations/MonitoringMetricsLogsAlerts
- Operations/ReliabilityAvailabilityRecoveryResilience
- Operations/MaintenanceUpdatesRetrainingOptimization
Data3 Types
- Data/ManagementSourcesQualityGovernance
- Data/ProcessingCleaningTransformationValidation
- Data/LineageTrackingVersioningAudit
Governance3 Types
- Governance/SecurityAccessEncryptionProtection
- Governance/CompliancePoliciesAuditsControls
- Governance/RiskBiasPrivacyMonitoring
Costs3 Types
- Costs/ComputeGPUsSchedulingUtilization
- Costs/InferenceTokensCachingBatching
- Costs/BudgetsTeamsAlertsReports
