Course-Aligned AI Teaching Support for Transnational Engineering Education: Design Conjectures, System Architecture, and Formative Deployment Evidence

This paper presents the first design-based research (DBR) cycle for a Course-Aligned Teaching Assistant (CATA). This retrieval-augmented AI system addresses the support gap in transnational engineering education (TNE). Informed by self-regulated learning and the zone of proximal development, CATA employs chapter-scoped retrieval over instructor-annotated lecture materials to maintain alignment with course terminology and pedagogical intent. Four design conjectures address TNE challenges: content grounding, scaffolded explanation, temporal availability, and multimodal interaction. Formative evidence from a semester-long deployment (2,628 interaction requests) reveals sustained usage consistent with active independent study. Each conjecture is evaluated against available evidence, with gaps identified for subsequent cycles.