As students increasingly leverage Generative AI tools for understanding course material and to generate solutions to homework questions, educators are confronted with challenges ranging from engaging students within classrooms to accurate student assessment models. In contrast to proposing bans on AI tools, we propose to leverage these tools to implement a course-aligned intelligent tutor iTutor. Our iTutor can effectively generate responses to student queries, that are within the scope of the course content, instigate implicit motivation, and is calibrated to offer guided assistance, instead of direct solutions. This facilitates learning and promotes critical thinking for students. It also enables responsible use of AI, without compromising integrity of student work or the learning process. Quintessential to the architecture of itutor and similar generative AI tools, are Large Language Models and Prompt Design. In this paper, we evaluate various LLMs and Prompt designs for high-quality response generation from iTutor. Our results show that Mistral7B, an open source LLM and our prompt designs are ideal candidates to implement the itutor that generates high quality responses. Furthermore, we show that our iTutor is monumental in fostering deep learning by generating responses that are aligned with the fundamental learning objectives of knowledge and comprehension. Finally, we show that introducing iTutor as a course resource will give course staff opportunity for redirecting time and energy to instigate active and experiential learning in classrooms and promote student engagement with course content.