The UIUC CS Department Doctoral Consortium pairs PhD students with peers and faculty mentors across multiple rounds so that students receive feedback from as many new perspectives as possible. Scheduling these groups was previously done by hand, an error-prone and exponentially hard task. We formulate the assignment as an integer optimization problem in Python: hard constraints require each student in exactly one group per session, group sizes of 2-5 students with 1-2 faculty, and no student grouped with their own advisor, while soft objectives minimize repeat student pairings and same-institution pairings across sessions. Given the number of students, faculty, sessions, advisors, institutions, and user-specified weights, the model produces an optimal schedule in under two seconds, reducing the time required by more than 99% relative to manual assignment.
