Platform and MLOps engineers harden shared model hosting used by multiple city agencies. Which governance role focus is this?

Platform/MLOps roles securing shared municipal model hosting operations focus on the multi-tenant runway agencies use: isolation, secrets, scaling, monitoring, deployment mechanics, and hardening of the shared inference/hosting layer. When platform and MLOps engineers harden shared model hosting used by multiple city agencies, they reduce common-mode risk—one weak hosting control would otherwise expose many departmental models. That governance-aligned operational focus works because shared platforms multiply both efficiency and blast radius.

Platform/MLOps roles tuning municipal model accuracy and performance fail because tuning accuracy is a model-engineering task, not the platform/MLOps role of hardening the shared hosting layer where concentration risk is highest. Platform/MLOps roles approving sensitive datasets for model training fail because data approval is a data-governance duty, not the platform/MLOps focus on hardening the shared hosting layer for multi-agency isolation and least privilege. Platform/MLOps roles validating municipal model compliance and risk reports fail because compliance validation is a separate assurance function, not the platform/MLOps engineering focus on hardening the shared host.

Exam caveat: platform hardening is necessary but not sufficient for model-specific risks (prompt injection content, unfair outcomes); exams distinguish platform controls from application/model governance. Operational check: enforce per-agency isolation on the shared host, rotate credentials automatically, monitor cross-tenant access attempts, and require security baselines (network policy, logging, image provenance) before any agency model is onboarded.