A 0.694 disparate-impact ratio flagged the synthetic tenant-screening model. The audit was only the starting line; the harder job was finding an inspectable change before another housing decision. We built Equora to keep that second step reproducible. On its fixed pool of 9,000 synthetic applications, the baseline audit measures the disparity. Then a bounded search evaluates 480 predefined, facially neutral configurations while holding the overall approval share constant. The selected alternative raises the lowest group ratio from 0.694 to 0.875. Measured AUC moves from 0.7823 to 0.7788, a 0.0036 change shown in the demo as a 0.36% accuracy cost. That gives the review team concrete evidence: the 480 evaluated options, the rule used to select one, and the result retained for review. The 0.80 threshold is the demo's configured four-fifths policy gate, not a legal conclusion or certification. For teams that oversee tenant screening, who owns this tradeoff decision in practice: model risk, compliance, or counsel? #FairHousing #TenantScreening #ModelRisk #AICompliance