Lumynary: Interaction with Data using LLM and Streamlit

Abstract: The "Lumynary: Interaction with Data using LLM and Streamlit" project seeks to develop an interactive platform to boost user interaction with data using Streamlit and large language models. Lumynary offers four main features, automated CSV and Excel file analysis, interaction with PDF using Google Gemini LLM, food recipe creation from images, and conversational interaction with YouTube videos. It combines cutting-edge tools such as FAISS for vector storage, LangChain for text processing, PIL for image processing, PyPDF2 for PDFs, and YouTubeTranscriptAPI for video transcripts. The prototype showcases phenomenal advancements in how users can interact with unstructured and structured data, offering automated insights, contextualized document discussions, and innovative content ideas. Lumynary differs from the rest by offering AI tool access to nontechnical individuals with multimodal data processing in one package. This step speaks volumes for the tremendous possibility that LLMs hold for reshaping data analysis and multimedia interaction in different domains. Keywords: Large Language Model, Google Gemini, Streamlit, Data Analysis, Data Interaction, Python JEL Classification Number: O33, C88, D83, M15