This paper addresses the emerging need for AI-driven teaching assistants to deliver personalized, effective and responsive educational assistance. Although large language models (LLMs) like GPT-3, GPT-4 generate human-responsive text, these models suffer from domain accuracy. One potential solution was identified as the Retrieval-Augmented Generation (RAG). These problems can be addressed by fusing the generative capability of LLMs with modern retrievals that allow it to refer to well-curated knowledge bases in that domain, thus making the responses more accurate and appropriate. This study develops RAG as an AI teaching assistant. This reduces the gap between general knowledge and subject-specific expertise, thus providing information that is not only relevant to the context but also customized to the unique requirements of students.