This module puts the review framework into practice. Learners score a complete use-case intake form against all six Responsible AI principles, assign an overall proceed, mitigate, or reject decision, and document the required follow-up actions for the product team.

This module develops learners' ability to interpret a pre-run bias-scan report, identify statistically significant bias indicators such as disparate impact across demographic attributes, and escalate findings as a P1 ethics incident — the foundational skill for any AI governance lead responsible for ethical oversight of deployed models.

This module develops learners' ability to evaluate multiple bias mitigation strategies — including Feature Reweighting and post-processing approaches — assess each against ethical and operational criteria, and justify a preferred control in a format suitable for steering-committee review.

This module develops learners' ability to evaluate a third-party dataset against an Internal Data Ethics Checklist—assessing sourcing transparency, consent documentation, representational fairness, and licensing compliance—and to produce a documented approve or reject decision that can withstand governance review.

This module develops learners' ability to trace dataset provenance using lineage metadata—identifying the origin, transformation history, and consent status of each training data source—and flag assets that lack the documentation required for ethical use in model training.

This module develops learners' ability to evaluate vector-store embeddings against a four-tier data classification schema, assign the appropriate sensitivity label, and enforce the corresponding retention tag — producing a classified and tagged embedding index that meets policy and audit requirements.

This module develops learners' ability to interpret data-quality dashboard metrics — including feature drift and outlier detection — assess whether quality thresholds have been breached, and trigger data-steward remediation workflows with documented tickets that give stewards everything they need to act.

In this project, learners produce a portfolio-ready Ethical AI Review Package that consolidates all Responsible AI work completed throughout LC 2 into a single integrated governance artifact. The resulting deliverable reflects the type of end-to-end Responsible AI review documentation used by AI governance and risk teams when preparing an AI system for internal approval, audit readiness, or deployment sign-off. Learners are expected to demonstrate not only the completion of individual tasks, but also the ability to connect ethical AI evidence into a coherent governance decision narrative—one in which use-case risk assessment, bias mitigation controls, dataset ethics, and data governance practices work together to support a clear, defensible deployment recommendation.