Christina Andonov

Serving and customizing models in production often involves running expensive GPUs at high utilization, maintaining low latency as traffic spikes, and scaling to demand without over-provisioning. Increasingly teams are doing it on Amazon EKS.

AI is creating new opportunities, and organizations need speed-to-market, scalability, performance, and cost efficiency to bring ideas to life. Amazon EKS provides the trusted and flexible foundation for your AI journey as you build new experiences using AI inference, agents, and model fine-tuning.

While developers can independently build Kubernetes applications, configuring cloud infrastructure dependencies remain a bottleneck, often consuming weeks. The goal of encapsulating cloud infrastructure in APIs has been challenging, traditionally requiring engineers skilled in programming and cloud infrastructure.

Organizations using Kubernetes have created self-service tools for developers to manage containerized applications and are starting to use these same tools to provision AWS services such as Amazon RDS, Amazon SQS, and others.