Building complex AI-powered applications often requires that the system has to parse and process an output of an AI. Previously, Amazon Bedrock developers had to instruct the AI to print JSON or XML as the response in their model prompts, which was error-prone and often contained hallucinations. With the introduction of Bedrock Agents, developers can define schemas in the OpenAPI format and instruct their AI to use external APIs to retrieve the required data or to perform the requested operations.

The first generation of GPU instances was introduced by AWS back in 2010. The industry has changed dramatically since then and nowadays server-side GPUs are widely used for a broad range of tasks - from machine learning to media processing. In this article we’re going to build a server that can securely run graphical applications in isolated environments and leverage on the power of modern GPUs.

Building and operating a machine learning system is a process that consists of many components and involves multiple specialists with various skill sets. The iterative nature and constant evolution of such systems amplifies the complexity to the point that it could become challenging to handle it even for a large and experienced team. Luckily, with the right tools at hands, it is possible to streamline this journey. Let’s take a look on how such pipeline could be built from the ground up.

Computer vision in general and facial recognition in particular are rapidly growing fields of computer science that can be beneficial in or even revolutionize some areas, such as transportation or retail. In this article we’re going to look closely at how such system could be built.

Amazon has recently released ARM64 support for Lambda based on Graviton2 processors. They claim that the new architecture provides better price performance. Let’s take a look how it stands up against the classic x64 CPU in some real-world scenarios.

AWS Lambda is a mature, feature-rich computing platform. While it’s very straightforward and simple to use for backend developers, when it comes to the performance tuning, there are a few things to keep in mind.