MA algorithms have received extensive attention from scholars soon after they were proposed, and are rapidly developing in various fields such as the problem of project schedule and the optimal allocation of cloud resources. In this paper, firstly, to address the lack of performance of the ordinary MA algorithm, Kent chaotic mapping is used in the initialization position to improve the initial traversal ability, and Sigmoid function is introduced to optimize the step-size settings in different stages of the crawling process. Different test functions are chosen to compare the efficiency of the solution and verify the performance of the improved MA algorithm. Finally, the improved MA algorithm is applied to allocate human resources in a city called China Mobile and a multi-project environment. Searching for the optimal solution and solving for the optimal value is the strongest performance of the improved MA algorithm compared to the other three algorithms. The human resource balance metrics of a city China Mobile are optimized and the peak human resource demand reaches the project below 120, reducing the waste of resources by about 20 people. In a multi-project environment, where 4 employees perform four different tasks, the optimal solution was found, resulting in an improvement in profitability.