The paper presents the results of analyzing the properties of HR eco-system as an object of neural network prediction and control. The eco-system is analyzed as an interaction space of multiple agents possessing the properties of: 1) human resources: abilities, competence, labor behavior; 2) working conditions expressed in complexity, labor tension; 3) certain quality of human resource management (HRM). Authors considered the possibility of describing the behavior of HR ecosystem and properties of its objects by models in the form of relatively finite state operating automata. This approach allows predicting the states of HR ecosystem and managing them with continuous learning. A new model of human resource management as a function of the properties of the ecosystem environment and HRM is presented. An optimization model is proposed to ensure balance and consistency of the HR ecosystem. Recommendations are given for using the obtained results in neural network prediction and management of HR ecosystem properties.