With the rapid expansion of artificial intelligence (AI) technologies and the changing expectations of the engineering labor market, revisiting university teaching and learning approaches has become increasingly necessary. This study aims to identify the antecedents, processes, and outcomes of teaching AI thinking to foster engineering students’ employability skills, using a theory-building meta-synthesis approach. A qualitative, inductive meta-synthesis was conducted based on the seven-step Finfgeld-Connett model. A systematic literature review was performed through structured searches in major Persian and international databases using guiding keywords. Studies were initially retrieved based on title–keyword relevance and then screened; ultimately, 50 articles were included for analysis and component extraction. Qualitative data were synthesized through open, axial, and selective coding, and the resulting model was subjected to descriptive, interpretive, theoretical, and pragmatic validation. The findings indicated that antecedents can be conceptualized as the knowledge, skills, and attitudes required of both students and faculty members. The process dimension comprised instructional mechanisms for developing AI thinking skills. Outcomes were articulated as the development of hard and soft employability skills among engineering graduates, particularly through engagement with AI tools and platforms. The proposed model provides a strategic framework for redesigning engineering education to enhance graduates’ employability.