As organizations put artificial intelligence and machine learning (AI/ML) workloads into continuous development and production deployment, they need to have the same levels of manageability, speed, and accountability as regular software code. The popular way to deploy these workloads is Kubernetes, and the Kubeflow and KServe projects enable them there. Recent innovations like the Model Registry, ModelCars feature, and TrustyAI integrations in this ecosystem are delivering these improvements for users who rely on AI/ML. These, and other improvements, have made open source AI/ML ready for use in production. More improvements are coming in the future.
Better Model Management
AI/ML analyzes data and produces output using machine learning “models,” which consist of code, data, and tuning information. In 2023, the Kubeflow community identified a key requirement to have better ways of distributing tuned models across large Kubernetes clusters. Engineers working on Red Hat’s OpenShift AI agreed and started work on a new Kubeflow component, Model Registry.