Design and Development of an AI-Powered SAT Math Practice Platform: a Blended Approach Integrating Design Thinking and LLMOps

This study presents the development of an AI-powered SAT Math practice platform designed to support Thai high school students through personalized, adaptive learning. Combining Design Thinking and LLMOps, the platform addresses key student challenges—such as unclear questions, lack of adaptivity, and low motivation—uncovered through interviews and user research. A custom data pipeline integrates official SAT materials and localized content, supported by a dual-database architecture for structured storage and semantic retrieval. Fine-tuned LLMs generate real-time, SAT-aligned practice questions and feedback. Iterative prototyping guided by user feedback led to interface refinements and improved usability. The platform incorporates real-time monitoring of performance metrics—accuracy, engagement, and response time—to dynamically adapt content. Usability testing confirms improved readiness, motivation, and user satisfaction. Future work will expand the question bank, enhance offline accessibility, and evaluate long-term learning outcomes. This study demonstrates how integrating Design Thinking and LLMOps can create scalable, learner-centered solutions in standardized test preparation.