AI EVALUATION · SELECTED WORK
Selected AI-agent evaluation scenarios authored by Luis Rodrigues and adapted for public portfolio presentation. Each case preserves the same canonical proof unit: objective, realistic context, user request, inferred task, required evidence, expected outcome, and a reference successful execution.
A fleet incident evaluation built around distributed evidence, source authority, current stakeholder resolution, and strict inference boundaries.
A launch-readiness task built around authoritative CI evidence, cross-system dependencies, deterministic state changes, and ordered closeout.
A repeat-outage evaluation testing current-policy selection, missing-input clarification, consistent parameter propagation, and approval-gated fallback behavior.
A customs-hold evaluation where the environment explains the incident but does not contain the criterion required to choose the correct refiling variant.
A cross-functional offboarding evaluation combining policy, missing-input clarification, eligible-record selection, commission calculation, access revocation, and People Ops handoff.
A financial-operations evaluation testing current-policy selection, missing-amount clarification, existing-customer reuse, and timing constraints.