Effective management of academic staff is central to the productivity of higher education institutions, yet traditional Human Resource Management (HRM) in Nigerian universities remains manual, subjective, and inefficient. This study proposes a Machine Learning (ML)-based HRM framework to enhance performance evaluation, workload balancing, and promotion recommendations. Using the CRISP-DM methodology, datasets on teaching load, research output, administrative roles, and professional development were collected and preprocessed. Three supervised ML algorithms—Random Forest, Support Vector Machines (SVM), and XGBoost—were trained and evaluated. Results show that the framework improves accuracy and consistency in staff assessments while reducing biases, with XGBoost outperforming other models in workload optimization and promotion prediction. The framework also features a scalable web-based architecture built with Flask and SQLAlchemy, supporting real-time analytics for administrators. This study presents a novel ML-driven HRM framework that enhances efficiency, fairness, and decision-making in resource-constrained academic environments.