Resource allocation is a pivotal component of project management that directly impacts project results regarding cost, duration and quality. Conventional Project Management Information Systems (PMISs) utilise static resource allocation techniques, which are insufficient for managing dynamic project circumstances, including variable deadlines, unexpected resource deficiencies and changing project requirements. This research introduces a machine learning-driven dynamic resource reallocation approach aimed at rectifying the flaws of traditional PMIS. The suggested system utilises predictive modelling techniques such as Decision Trees, Random Forest and XGBoost to estimate task completion times, human availability and workload distribution, thereby enabling proactive resource allocation. Furthermore, human-centric elements, including employee well-being and job satisfaction, are included in the allocation process, alleviating stress and enhancing productivity. The system employs anomaly detection to recognise real-time imbalances and adjust resources accordingly. The results indicate that machine learning models markedly improve the efficiency and scalability of resource allocation in PMIS, facilitating enhanced project success.