Objectives:

The objective of the session was to introduce students to Retrieval-Augmented Generation (RAG) architecture and its application in building intelligent, context-aware AI assistants. The session aimed to bridge the gap between theoretical knowledge of large language models and their practical deployment using modern platforms. Another objective was to familiarize participants with Streamlit and Snowflake Cortex Search as tools for developing scalable and enterprise-ready AI applications.

Detailed Report:

The institution organized an expert session on “Build a RAG-Based LLM Assistant Using Streamlit and Snowflake Cortex Search,” delivered by Mr. Rahul Boorji, Principal Consultant – Gen AI at Genpact and an alumnus, who shared his technical expertise and industry experience.

Mr. Boorji began by explaining the limitations of standalone large language models and the need for Retrieval-Augmented Generation to improve accuracy, relevance, and reliability of AI outputs. He described how RAG systems combine document retrieval with generative models to produce context-aware responses.He then walked participants through the architecture of a RAG- based assistant, outlining the role of data ingestion, vector storage, embedding generation, retrieval mechanisms, and response generation. He demonstrated how Streamlit can be used to build an interactive front-end interface and how Snowflake Cortex Search enables efficient, secure, and scalable retrieval of enterprise data.The session included a high-level demonstration of how documents are indexed, searched, and integrated with LLMs to create a functional assistant. Mr. Boorji also discussed considerations such as data privacy, security, model evaluation, and responsible AI practices.

The session concluded with an interactive Q&A, encouraging students to explore practical AI development and apply these technologies in real-world business and research contexts.

Learning Outcomes:

Participants gained an understanding of RAG architecture and its role in enhancing the performance and reliability of LLM-based systems. They learned the functional components of a RAG pipeline and the use of Streamlit and Snowflake Cortex Search in building AI applications. The session enhanced their awareness of enterprise AI deployment, data governance, and responsible AI practices. Overall, participants developed foundational knowledge to begin experimenting with and implementing RAG-based AI solutions.

Prepared By:

Mr. Rahul Boorji