Under CoE standards, a data scientist prepares labeled datasets for the housing-authority fraud model. Which role responsibility is illustrated?

Data scientist work preparing labeled datasets for civic AI models is the role responsibility shown when, under CoE standards, a data scientist prepares labeled data for a housing-authority fraud model. That role focuses on dataset design, labeling quality, bias-aware sampling, documentation, and collaboration with domain experts so supervised learning has trustworthy inputs. It fits municipal CoE practice because model quality and fairness start with how housing-related examples are selected and labeled—not only with the training algorithm.

Data scientist work approving labeled datasets for civic AI models fails because approving labeled datasets is a governance sign-off function, not the hands-on dataset-preparation work a data scientist role is being tested on here. Data scientist work auditing labeled datasets for civic AI models fails because independent assurance is an auditor function; data scientists building training sets are not providing independent audit of those datasets. Data scientist work deploying labeled datasets for civic AI models fails because deployment is an ML engineering/operations responsibility, not the labeled-dataset preparation a data scientist role is being tested on here.

Exam caveat: data scientists may support risk discussions, but ownership of dataset approval, independent audit, and deployment decisions remains outside the preparation role; exam questions often test role boundaries more than tooling. Operational check: under CoE standards, require dataset cards (source, labeling guide, known gaps) for the housing-authority fraud model, peer-review label schemas with program staff, and segregate production decision authority from dataset preparation duties.