Cohort Institute

The problem

Instructors who use team projects form the teams by hand, or at random, or by letting students pick. Each method has a known failure: hand-forming doesn't scale past a few sections, random ignores everything that matters, and self-selection reproduces existing friend groups.

Eddie's role

Founder and builder. Product design, matching logic, and the instructor-facing workflow.

Stage

Pilot with a university partner. In use by instructors this term.

What informed the approach

Team formation research on skill diversity and schedule compatibility. Instructor interviews about what they actually optimize for (mostly: avoid the team that falls apart in week three). Constraints: must work from data instructors already have, must be explainable when a student asks why they were placed.

What was built

An instructor workflow: define the class, collect a short student survey, set what matters for this project, generate teams, review and adjust, publish. Matching balances the factors the instructor selected and shows why each team was formed.

What is known so far

Instructors can form teams for a full section in minutes instead of hours. Whether the teams are better is the open question.

Limitations and next evaluation

"Better teamwork" has no single metric. Next step is agreeing with instructors on what to compare (peer evaluations, project outcomes, drop rate) before claiming the tool improves anything.