This paper presents a semi-automatic approach to assigning students to project teams for a year-long, industry-sponsored senior capstone course. Successful assignment requires knowl- edge of at least individual project requirements, student skills, student personalities, and student project preferences. This mix of hard skills, soft skills, and interpersonal impres- sions requires human involvement to produce a high-quality assignment. The importance of faculty input often requires that the assignment process be labor- and time-intensive. Our approach attempts to reduce the time required to perform this assignment by selectively automating parts of the task flow. An automated search uses a randomized greedy algorithm combined with local optimizations to explore a large space of solutions. Candidate “good” solutions are then presented to capstone faculty. Criteria such as skill set, student capability, and personality compatibility are applied by human evaluators to reduce the candidate solution set. These candidate solutions are then distributed to small groups of faculty to look for improvements using system-generated tables of options. This approach leverages automation at appropriate stages while keeping the experts—the faculty—involved in the selection process. Our initial implementation has reduced the time needed to select an allocation by about a factor of three over previous manual approaches.