Introduction

NPO Model Management is an end-to-end automation platform for AI/ML models promoted by SKAX, designed to manage the entire lifecycle—including development, registration, deployment, testing, and operation—in a single integrated environment. Users can register models in a central Model Registry and deploy them to Kubernetes clusters via a browser-based interface, without the need for complex infrastructure setup or manual processes. After deployment, users can perform real-time testing and performance validation in the Playground. During the operation phase, the Monitoring & Management feature enables real-time monitoring of model status, resource usage, and logs, allowing for rapid troubleshooting. This automated workflow minimizes human intervention (No People Operation), maximizing efficiency, stability, and reproducibility in AI model operations.

NPO Model Management consists of the following key features:

• Model Registry: Model registration, version control, and metadata management
• Model Serving: Kubernetes-based model deployment and service operation
• Playground: Real-time testing and validation of deployed models
• Monitoring & Management: Monitoring and controlling the status of deployed models

Through these integrated features, NPO provides an MLOps-based standard platform for model management, enabling transition of models from research and development to production environments.