Аннотация: In the process of modeling organizational and technological systems, a lot of problems arise due to the imperfection of classical algorithms. In particular, in the process of placing components according to certain metrics, such algorithms are very slow, since in the ideal case they need to go through all possible options in order to find the optimal solution. With a high number of components that make up the systems, this can lead to several days of simulation. The use of machine learning is designed to reduce this time while maintaining the quality of the resulting models of organizational and technological systems.