This section lists and introduces the blogs we have posted on AWS Study Group.
Modern applications increasingly require large memory capacity to process big data, perform complex analytics, or run Machine Learning (ML) models. To address this challenge, AWS introduced AWS Lambda Managed Instances — a solution that overcomes the memory limits of standard Lambda while retaining the full convenience of a Serverless architecture.
AI agent applications have a very unique traffic pattern: an agent can fire hundreds of vector queries during a single task inference, then remain idle for hours. With traditional search infrastructure, you still pay for unused capacity during those idle periods. On May 28, 2026, AWS announced the next generation of Amazon OpenSearch Serverless (called NextGen) — a search and vector engine redesigned from the ground up to serve exactly this type of unpredictable workload.
Modern applications often need to handle multi-step sequential processes — such as order processing, user onboarding, or orchestrating AI workflows that call Language Models (LLMs) and external tools multiple times. Previously, to accomplish this on AWS Lambda, developers had to piece together Step Functions, SQS, and DynamoDB for state management, leading to complex infrastructure. At re:Invent 2025, AWS introduced AWS Lambda Durable Functions — enabling you to write entire workflows as sequential code within a single Lambda function while maintaining fault tolerance and long-term pause capabilities.