In this paper, we construct a multi-level network based on corporate mobility relationships to quantify human resource attributes. The cuckoo search algorithm (CS) is chosen to enhance the global optimization capability of human resource management scheme. Combine CS and XGBoost to construct CS-XGBoost algorithm, and realize the optimal solution of HRM scheme through hyperparameter optimization and other steps. The multi-project human resource management of construction enterprises is taken as an example to verify the auxiliary value of CS-XGBoost algorithm in the generation of optimal management scheduling scheme. Empirical studies show that the algorithm can obtain the optimal solution in about 450 iterations. In multi-project scheduling management, the optimal duration can be reduced to 510 days, which is better than the comparison algorithm. With the introduction of demand prioritization requirements, the algorithm can effectively balance the differences in project duration, project cost and employee working time. The CS-XGBoost algorithm can be used to quickly realize the optimal decision-making of enterprise human resource scheduling management, save costs and improve efficiency.