AI and ML workloads in Azure Kubernetes Service (AKS)

This article provides an overview of running artificial intelligence (AI) and machine learning (ML) workloads in Azure Kubernetes Service (AKS).

AI and ML today

Artificial intelligence (AI) and machine learning (ML) are transforming the way we solve complex problems and deliver value to customers. Generative AI has produced a paradigm shift that has transformed applications into data-driven, personalized experiences across a full range of applications involving text, code, images, videos, voice, music, and more. We've also seen an increase in model size, operational complexity, and an increased need for security and cost-efficiency.

To develop cutting-edge AI models, you need to be able to allocate computing resources across diverse workloads. This includes model training, serving, inference, and managing auxiliary tasks across infrastructure and workflow orchestration.

It's essential to enable easy integration with open-source software, frameworks, and data platforms. Developing, refining, optimizing, deploying, and monitoring ML models can be challenging and complex. These inherent complexities require a platform, like AKS, that allows you to maximize the performance of your AI and ML workloads while reducing inefficiencies and bottlenecks.

Develop and deploy AI and ML applications with AKS

AKS is an ideal platform for deploying and managing containerized applications that require high availability, scalability, and portability through the use of open-source tools and integration with existing DevOps processes.

Using AKS to host your AI and ML applications enables you to leverage high-performance infrastructure, cost-efficiencies, and robust security measures while maintaining the pace of innovation. AKS reduces the operational burden of running ML workloads that are optimized for the computational and infrastructure load induced by developing and deploying AI applications.

Design and deploy AI and ML workloads on Azure

• Deploy an application that uses OpenAI on Azure Kubernetes Service (AKS)
• Deploy an AI model on Azure Kubernetes Service (AKS) with the AI toolchain operator add-on
• Configure and deploy a Ray cluster to accelerate ML workloads on Azure Kubernetes Service (AKS)
• Build and deploy data and machine learning pipelines with Flyte on Azure Kubernetes Service (AKS)

Contributors

Microsoft maintains this article. The following contributors originally wrote it:

• Colin Mixon | Product Manager
• Erin Schaffer | Content Developer 2
• Brian Redmond | Principal PDM Manager
• Ken Kilty | Principal TPM
• Russell de Pina | Principal TPM