MLOps, Models, and Responsible AI

This unit deepens learners’ understanding of foundation models, the transformer architecture, and the relationships among AI, ML, and deep learning while introducing practical MLOps pipelines and AWS AI services (Bedrock, SageMaker). Students will learn model customization, evaluation, tokenization/embeddings, inference parameter tuning, and apply governance, security, and responsible-AI practices to prepare models for production and subsequent advanced implementation work.

Learning Objectives

• Analyze and compare foundation model and transformer concepts and explain the relationships among AI, ML, and deep learning to inform model customization and evaluation decisions
• Apply AWS AI services (e.g., Amazon Bedrock, SageMaker) to construct and configure an MLOps pipeline that incorporates tokenization, chunking, embeddings, knowledge bases, and inference-parameter tuning
• Demonstrate responsible AI implementation by configuring security controls, governance processes, compliance checks, and bias-mitigation practices using AWS tools and policies
• Evaluate model performance and suitability by conducting customization and evaluation experiments, selecting appropriate metrics, identifying use cases unsuitable for AI/ML, and recommending operational improvements