Run:AI provides a fully supported pipeline orchestration solution, powered by Kubeflow Pipelines, the leading open-source framework for ML pipelines. Based on Kubernetes, Run:AI provides a managed solution for running ML pipelines in the simplest, most efficient way, whether in the cloud, on-premises or at the edge.
Automation in production is highly important for creating standardization, deployment procedures, agile development, and more. For Machine Learning in production, automation relates also to running pipelines, a sequence of tasks with dependencies between them. From data engineering pipelines, retraining models in production to running multiple inference models on a batch of collected data, ML pipelines become a necessity for any ML team.
Running pipelines can be cumbersome. Each pipeline launches multiple tasks with different resource requirements, in parallel or in sequence, where all tasks share data and resources. Caching and dynamic resource allocation, including freeing up and quickly provisioning compute instances, can be highly important for efficient pipeline execution.
Learn more from our product documentation >>