II.
Workflow overview
Reference · liveworkflow:model-training-pipeline
Model Training Pipeline overview
End-to-end ML model training workflow — data preparation, feature engineering, hyperparameter search, distributed training, evaluation against baselines, and model registry promotion. Includes reproducibility checks and experiment tracking. Distinct from model-training-cycle in focusing on pipeline infrastructure rather than individual training runs.
Attributes
displayName
Model Training Pipeline
workflowKind
development
triggerType
event-driven
typicalCadence
per-experiment
complexity
cross-team
description
End-to-end ML model training workflow — data preparation, feature
engineering, hyperparameter search, distributed training, evaluation
against baselines, and model registry promotion. Includes
reproducibility checks and experiment tracking. Distinct from
model-training-cycle in focusing on pipeline infrastructure rather
than individual training runs.
Outgoing edges
applies_to_domain2
- domain:ml-ai·DomainML/AI
- domain:ml-ops·DomainMLOps
involves_role3
- role:ml-engineer·RoleMachine Learning Engineer
- role:applied-scientist·RoleApplied Scientist
- role:data-engineer·RoleData Engineer
requires_skill_area3
- skill-area:machine-learning-frameworks·SkillAreaMachine Learning Frameworks
- skill-area:feature-engineering-pipelines·SkillAreaData and Feature Engineering Pipelines
- skill-area:ci-cd-ml-pipelines·SkillAreaCI/CD for ML Pipelines
triggers_responsibility2
- responsibility:model-training-quality·ResponsibilityModel training quality
- responsibility:model-quality-assurance·ResponsibilityModel quality assurance
Incoming edges
None.