To reduce isolation, the literature recommends grouping two or more female or minority students together when group work is performed in courses where they are underrepresented, but it can be time consuming for instructors to hand select ideal groups. GroupEng – an open-source program available to all educators – was created to make it easy for instructors to use variables such as gender, race, student performance (e.g. early test scores), and student interests to auto generate balanced groups in minutes. This paper by the creators of GroupEng talks about the research behind the program, compares the use of GroupEng with hand selected groups in real courses, and explains how any instructor can use the program in their STEM course.