Bedrock Generative Model Workflows

This unit trains students to integrate and operate generative AI models within the Amazon Bedrock ecosystem, covering image generation, multimodal context, safety guardrails, parameter tuning, and end-to-end pipelines. Students will practice building text-to-SQL systems, maintaining conversational context, and leveraging SageMaker JumpStart for model instantiation and deployment readiness.

Learning Objectives

• Analyze multimodal use cases and select appropriate Amazon Bedrock foundational models (image, text, and multimodal) based on data characteristics and task requirements
• Apply Stable Diffusion and multimodal model techniques in Amazon Bedrock by configuring prompts and model parameters to generate and iteratively refine high-quality images and contextual outputs
• Demonstrate implementation of Bedrock guardrails and context-management strategies—including system messages, session state handling, and parameter controls—to maintain safe, coherent, and persistent chatbot behavior
• Construct and evaluate an end-to-end text-to-SQL pipeline using Amazon Bedrock and integrate Amazon SageMaker JumpStart components, measuring query accuracy, robustness, and overall system performance