Designing a Merit-Based Recruitment Model in Bank Saderat Using Artificial Intelligence

Abstract

This study aims to design and validate an optimal merit-based recruitment model for Bank Saderat using artificial intelligence techniques. A mixed-methods design was employed. In the qualitative phase, a Delphi approach involving human resource managers and experts of Bank Saderat was used to identify and finalize competency components. In the quantitative phase, employees’ performance, test scores, job-related, personal, and socio-cultural data were collected, preprocessed, and modeled using four machine learning algorithms: Deep Neural Network, Ridge Regression, XGBoost, and Support Vector Machine. Adjusted R² and mean absolute error were applied to compare predictive performance. The results indicated that the Deep Neural Network achieved the highest predictive power among all models. Across all algorithms, teamwork, communication skills, customer orientation, and organizational commitment were consistently identified as the most significant predictors of employee competency, whereas variables such as ethnicity, political orientation, place of birth, and workplace location showed minimal predictive importance. The findings demonstrate that artificial intelligence–based models can enhance merit-based recruitment by prioritizing performance-related competencies and minimizing irrelevant biases, thereby supporting more objective and efficient human resource decision-making in the banking sector.