COMPEL Glossary / model-lifecycle-management
Model Lifecycle Management
Model lifecycle management is the governance discipline of maintaining visibility, control, and accountability over AI models from initial conception through production deployment, monitoring, retraining, and eventual retirement.
What this means in practice
It recognizes that AI models are not static assets but living systems that degrade, evolve, and influence decisions every day they operate. Key lifecycle stages include development (model design and training), validation (independent performance and fairness assessment), deployment (production release with governance controls), monitoring (continuous performance and drift tracking), retraining (updating models when performance degrades), and retirement (decommissioning models that no longer meet standards). In the COMPEL framework, model lifecycle management is the intersection of MLOps (Domain 7) and AI Governance Structure (Domain 18), requiring both technical infrastructure and governance processes.
Why it matters
AI models require governance through their entire lifecycle: development, validation, deployment, monitoring, retraining, and retirement. Organizations that focus governance only on initial deployment miss the ongoing risks of models degrading in production, being retrained on inappropriate data, or remaining active long after they should have been retired. Lifecycle governance ensures continuous oversight rather than one-time approval.
How COMPEL uses it
Model lifecycle management sits at the intersection of MLOps (Domain 7) and AI Governance Structure (Domain 18), requiring both technical infrastructure and governance processes. During Calibrate, lifecycle management maturity is assessed. The Model stage designs lifecycle governance policies. The Produce stage implements lifecycle tracking through model registries and monitoring. The Evaluate stage audits lifecycle compliance and retirement readiness.
Related Terms
Other glossary terms mentioned in this entry's definition and context.