Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn to build sustainable and cost-effective generative AI solutions on AWS in this 20-minute conference talk that addresses the environmental impact of AI development. Explore the energy challenges posed by growing model sizes and increasing data center demand, then discover how AWS reduces carbon footprint through renewable energy and efficient data centers. Master the fundamentals of sustainable AI architecture by starting with managed and serverless services including Amazon Bedrock, EKS, and SageMaker. Understand strategic model selection principles to avoid using unnecessarily large language models and leverage Amazon Bedrock Evaluations for proper model assessment. Compare different customization approaches including prompting, retrieval-augmented generation (RAG), fine-tuning, and training from scratch to determine the most efficient option for your use case. Optimize your infrastructure by choosing appropriate silicon solutions with AWS Trainium for training workloads and Graviton processors for inference tasks. Implement advanced inference optimization techniques through model compression, compilation, and distillation to reduce computational overhead. Establish comprehensive observability and guardrails to monitor cost, performance metrics, and responsible AI practices throughout your GenAI pipeline.
