We demonstrate the application of machine learning methods for predicting the properties of epitaxial structures in multi-parameter technological processes characterized by complex nonlinear dependencies. The synthesis of silicon carbide thin films on silicon substrates via atomic substitution method was investigated as a model system. A neural network model capable of predicting key characteristics of the resulting SiC films based on synthesis process parameters, including pressure, temperature, substrate type, and other additional synthesis conditions, was developed. Comprehensive optimization of the model architecture was performed followed by validation of prediction accuracy. The high efficiency of machine learning algorithms for analyzing and controlling complex epitaxial processes was demonstrated. Keywords: machine learning, neural network model, epitaxial growth, SiC, Si, atomic substitution method.