Maya V. Bernavskaya

The digital transformation of higher education and the broad implementation of project-based learning require effective educational interaction models to overcome traditional barriers such as lack of individual support, low student autonomy at undergraduate level, limited thematic diversity of projects, and psychological barriers in instructor communication. The study tests and evaluates the effectiveness of an innovative model of facilitated interaction between graduate and undergraduate students integrated with neural network technologies in the "Project Activities" course. Competence-based, facilitation-oriented, and environmental approaches, along with peer-to-peer learning principles, form the methodological basis. The pedagogical experiment used normative and comparative analysis, educational process design and modeling, surveys, testing, expert evaluations, and comparative statistical analysis. Tested within the 19.03.01 "Biotechnology" program at Peter the Great St. Petersburg Polytechnic University, the model significantly increased student autonomy and motivation, shortened project stages, improved satisfaction ratings (from 2.7 to 4.2 on a 5-point scale), and developed sustainable pedagogical and project-management competences among graduate student moderators (94% reported growth in facilitation and group dynamics skills). The work presents a comprehensive structural-functional educational interaction model combining horizontal mentoring and digital tools advantages. The model fosters targeted development of professional and transversal competences (soft skills) for all participants, that is supported empirically. Developed guidelines and neural network tools are ready for implementation in other universities, representing a promising direction for digital transformation of project-based learning.