Purpose Develop a simple local AI workflow using the Retrieval-Augmented Generation (RAG) AI design pattern Experiment with multiple open-weight AI models and assess performance Gain exposure and hands-on experience with the latest AI tools Build RAG-Based LLM App Takes PDF document as input and allows user to ask questions about the document via chat Tech Stack AI Framework: LangFlow LLM Model: Gemma 4 + Ollama Vector Database: DataStax Astra DB Architecture Flow Sample Chat Input / Output Insights Using model gemma4:latest over qwen3.5:latest improved response times significantly Average of 50s for qwen3.5 verseus average of 20s for gemma4 Local Deployment Clone Git repository in local environment Execute: docker compose up Access local instance of Langflow at: http://localhost:7860
