With the introduction of Large Language Models (LLMs), chatbots have been increasingly utilized in various fields. One such field is education, where chatbots can offer personalized tutoring services on demand. They can handle subject-specific questions, explain complex concepts, provide feedback on practice exercises, and assist students in navigating learning materials. However, ensuring that the chatbot delivers accurate and valuable information in a specific domain necessitates substantial time and resources to customize the pre-trained model. This paper presents a hybrid approach to developing a chatbot capable of addressing user inquiries regarding specific course materials. Our chatbot employs LLMs for contextual question answering and integrates the Retrieval Augmented Generation (RAG) model for domain-specific queries. We train an embedding-based retrieval model using ChromaDB and BM25Retriever. The LLM employed is the Google Gemini 1.0 Pro model, accessed via the Gemini API. Experimental outcomes demonstrate that our chatbot outperforms well-known general-purpose public chatbots in terms of retrieval accuracy and the capacity to identify out-of-domain queries. This chatbot serves as a valuable resource, providing prompt and convenient information, thereby reducing the time spent searching through source documents. In the future, it has the potential to enhance the learning experience and support educators in the realm of human-machine collaboration in education.