Blog_dumb

Foundation model training has reached an inflection point where traditional checkpoint-based recovery methods are becoming a bottleneck to efficiency and cost-effectiveness. As models grow to trillions of parameters and training clusters expand to thousands of AI accelerators, even minor disruptions can result in significant costs and delays. In this post, we introduce checkpointless training on …

Checkpointless training on Amazon SageMaker HyperPod: Production-scale training with faster fault recovery Read More »

Modern AI infrastructure serves multiple concurrent workloads on the same cluster, from foundation model (FM) pre-training and fine-tuning to production inference and evaluation. In this shared environment, the demands for AI accelerators fluctuates continuously as inference workloads scale with traffic patterns, and experiments complete and release resources. Despite this dynamic availability of AI accelerators, traditional …

Adaptive infrastructure for foundation model training with elastic training on SageMaker HyperPod Read More »

Large Language Model (LLM) agents have revolutionized how we approach complex, multi-step tasks by combining the reasoning capabilities of foundation models with specialized tools and domain expertise. While single-agent systems using frameworks like ReAct work well for straightforward tasks, real-world challenges often require multiple specialized agents working in coordination. Think about planning a business trip: …

Customize agent workflows with advanced orchestration techniques using Strands Agents Read More »

Enterprise organizations are rapidly moving beyond generative AI experiments to production deployments and complex agentic AI solutions, facing new challenges in scaling, security, governance, and operational efficiency. This blog post series introduces generative AI operations (GenAIOps), the application of DevOps principles to generative AI solutions, and demonstrates how to implement it for applications powered by …

Operationalize generative AI workloads and scale to hundreds of use cases with Amazon Bedrock – Part 1: GenAIOps Read More »

Amazon Simple Storage Service (Amazon S3) is a highly elastic service that automatically scales with application demand, offering the high throughput performance required for modern ML workloads. High-performance client connectors such as the Amazon S3 Connector for PyTorch and Mountpoint for Amazon S3 provide native S3 integration in training pipelines without dealing directly with the …

Applying data loading best practices for ML training with Amazon S3 clients Read More »

We’re sharing a practical playbook to help organizations streamline and enhance sustainability reporting with AI.Corporate transparency is essential, but navigating frag… ​ 

As cloud infrastructure becomes increasingly complex, the need for intuitive and efficient management interfaces has never been greater. Traditional command-line interfaces (CLI) and web consoles, while powerful, can create barriers to quick decision-making and operational efficiency. What if you could speak to your AWS infrastructure and get immediate, intelligent responses? In this post, we explore …

Building a voice-driven AWS assistant with Amazon Nova Sonic Read More »

When you go Live with Search, you can have a back-and-forth voice conversation in AI Mode to get real-time help and quickly find relevant sites across the web. And now, … ​ 

We’re bringing Gemini’s state-of-the-art translation model to Google Translate for text, and more new features. ​ 

Gradient Canvas is a new art exhibition celebrating a decade of creative collaborations between artists and artificial intelligence. ​