Large Language Models (LLMs) are transforming many fields, but their limitations regarding factual accuracy and access to specific, up-to-date information are well-known. Retrieval-Augmented Generation (RAG) provides a powerful framework to address these challenges by grounding LLMs with relevant external knowledge. This session offers a comprehensive journey into RAG, divided into two parts: First, a 30-minute talk will demystify RAG concepts, exploring the core architecture (retriever + generator), illustrating its benefits, and showcasing practical demonstrations in academic contexts like literature search and Q&A over course materials. Following this, a 30-minute hands-on tutorial (using Google Colab) will guide participants step-by-step to build their own basic RAG pipeline. You will learn to index a small dataset, implement a simple retrieval mechanism (e.g., BM25), and integrate an LLM API to generate context-aware answers. Attendees will leave with both a solid conceptual grasp of RAG and the practical ability to start building their own grounded AI applications. Basic Python familiarity is recommended. Bring your laptop to participate in the tutorial