Challenge-Based Learning (CBL) is an educational methodology in which students work in teams to address real-world problems and develop prototype solutions, guided by a coach. CBL offers various benefits, including the development of technical knowledge—depending on the challenge theme—and soft skills such as teamwork, problem-solving, and communication in multicultural settings. However, in learning and innovation environments, participants' experiences are strongly influenced by their interactions with teammates. Previously, an AI-driven team formation algorithm has been used to create teams based on individuals’ skills, personalities, and interests. This internship thesis proposes the use of the AI team formation algorithm Edu2Com to assess its accuracy in predicting team effectiveness within the context of the SDG Olympiad competition, where teams are evaluated by a jury on their final projects. Additionally, students will evaluate their own team's effectiveness. This thesis contributes to the literature on team composition in CBL settings and offers practical recommendations for future implementations of the SDG Olympiad.