Russian University Deploys AI to Match Students With Academic Advisors
At Immanuel Kant Baltic Federal University, a new machine learning system helps students and PhD candidates quickly find the right mentor by analyzing research interests, skills, and compatibility.
Researchers at Immanuel Kant Baltic Federal University have developed a system designed to save time and streamline the process of finding an academic advisor.
Information about potential supervisors has been encoded into machine learning algorithms. When students fill out their profiles, the system generates a ranked list of faculty with whom they are most likely to have productive collaboration.
The developers tested two approaches: a fast one and a more precise one. In the instant mode, the algorithm ranks all candidates by similarity to the student’s profile and then refines the results. In the precise mode, the system compiles a shortlist of 10 potential supervisors based on research fields and academic schools, checking each match against three criteria: interests, skills, and competencies.
The platform is built on modern neural networks (Qwen3 and Falcon), the Transformers library, PyTorch, and FastAPI. Data on faculty supervisors was collected from the university’s website, converted into JSON format, and encoded into vector representations for semantic search.
