In this talk, I will share my experience developing and deploying a customized AI-powered teaching assistant for Math 359: Mathematical Theory of Data Science. The tutor is available 24/7 through a user-friendly web interface, built with the OpenAI API, and aligned with course materials, including lecture notes, practice problems, homework, and quizzes, to deliver course-specific support. Rather than providing direct solutions, the AI tutor is designed to coach students through structured explanations, targeted examples, and strategic hints. The goal is to promote productive struggle, strengthen conceptual understanding, and support independent problem-solving while maintaining academic integrity.