Human Resour (HR) departments are leveraging data-driven decision-making for more effective management of absenteeism, promotions, and retirement planning. Conventional methods are often slow and subjective, especially in big companies like oil firms. In this research, we propose an AI-enhanced Human Resource Management System (AI-HRMS) which consolidates absenteeism forecasting, promotion prediction and retirement risk estimation into one decision-making assistance framework. Using actual HR data from the oil industry, we pre-process this raw data and build the physical world models using Random Forest, CatBoost and Artificial Neural Networks (ANN). SMOTE or under-sampling techniques were used to balance the datasets based on the type of prediction task. All models were evaluated with accuracy, precision, recall, F1-score, ROC-AUC and PR-AUC. The models are deployed as standalone AI services in a Docker container and seamlessly integrated into HRMS for real-time inference. Overall performance was excellent for the identification of absenteeism risk, prediction on retirement, and estimation of promotion eligibility for all models. The integration was used to provide real-time decision support to HR managers. After all, use of AI in supporting HR planning through the multi-model AI-HRMS system supports reduction of manual work and is able to maintain consistency in large organization decision making and help establish a modular architecture that is future-ready for integration with a host of HR analytics systems.