International Journal of Scientific Research and Engineering Development

This study develops and evaluates ethical machine learning frameworks designed to detect and mitigate bias in automated hiring systems. Employing a mixed-methods approach, we conducted comparative analyses of prominent frameworks—AI Fairness 360 (AIF360) and Fairlearn—on recruitment datasets, complemented by qualitative interviews with 82 HR professionals. Our evaluation utilized modified UCI Adult Income data and synthetic hiring records, applying demographic parity and equalized odds metrics. Key findings reveal that AIF360 detected 60% more subtle biases than baseline models, while hybrid framework implementations achieved a 45% reduction in demographic disparities across protected attributes. Integrated auditing mechanisms improved hiring equity scores by 35% and enhanced transparency through counterfactual explanations. Results demonstrate that combining preprocessing bias mitigation with continuous monitoring dashboards significantly advances fairness objectives. However, implementation challenges persist, including regulatory ambiguity and computational overhead. This research underscores the necessity of mandatory bias audits in HR technology certifications and proposes actionable strategies for equitable recruitment practices in increasingly automated labor markets.