Ml Model Selection

Trigger Boundary

• •Use when ML data, model, training, evaluation, or serving choices are being made.
• •Do not use for generic API lifecycle governance; use api-*.
• •Do not use for non-ML database administration concerns.

Goal

Produce reliable ML lifecycle decisions from data to production monitoring.

Inputs

• •Change scope and risk profile
• •Domain evidence for candidate model trade-offs across accuracy, latency, and cost
• •Operational, compliance, and rollout constraints

Outputs

• •Model selection decision log
• •Decision log for candidate model trade-offs across accuracy, latency, and cost
• •Verification checklist with measurable pass-fail criteria

Workflow

1. •Clarify outcomes and hard constraints for candidate model trade-offs across accuracy, latency, and cost.
2. •Produce options and select an approach for candidate model trade-offs across accuracy, latency, and cost.
3. •Evaluate trade-offs across security, performance, operability, and maintainability.
4. •Verify decisions using head-to-head benchmark and robustness comparison.
5. •Publish decisions, residual risks, and accountable follow-up actions.

Quality Gates

• •Scope and assumptions for candidate model trade-offs across accuracy, latency, and cost are explicit and reviewable.
• •Decision rationale is backed by evidence instead of preference.
• •Rollout and rollback criteria are defined when production impact exists.
• •Residual risks have owners, due dates, and verification steps.

Failure Handling

• •Stop when selected model lacks evidence against viable alternatives.
• •Escalate when accepted risk exceeds team policy thresholds.