Amazon SageMaker AI Pipelines

license

Proprietary

service

proprietary(AWS managed service)

source

closed(no implementation is published

A managed AWS service. AWS publishes a Python SDK that authors and submits pipelines, but the orchestration service itself is unpublished, and AWS states it provisions and manages the orchestration compute on the user's behalf. Nothing about a license can help where there is no self-hostable implementation, which is why the ladder settles this on source alone.

• https://docs.aws.amazon.com/sagemaker/latest/dg/pipelines.html recorded 2026-09-15

  AWS's own documentation: 'Amazon SageMaker Pipelines is a purpose-built workflow orchestration service to automate machine learning development', with 'auto-scaling serverless infrastructure - you don't need to manage the underlying orchestration infrastructure to run Pipelines, which allows you to focus on core ML tasks. SageMaker AI automatically provisions, scales, and shuts down the pipeline orchestration compute resources.' Establishes a managed service with no published implementation.

No usage figure is published for this service specifically. The vendor's platform as a whole is widely used, but nothing disclosed is specific to the orchestration product being scored, and it publishes no countable artifact - no package, repository or registry entry - so no level is assigned rather than one being inferred from the parent platform.

• https://aws.amazon.com/sagemaker/ai/customers/ recorded 2026-09-15

  AWS's SageMaker AI customers page, read for a usage disclosure specific to Pipelines; it names customers but publishes no count of pipelines, runs or customers using the orchestration service.

Admitted on the ML-job-model limb: training and tuning steps are first-class types it schedules and tracks. Band 3, level with the other managed pipeline services. It was drafted at 4 on the reading that Pipelines selects each step's hardware, and the cited documentation does not show that - it shows AWS provisioning the ORCHESTRATION compute. The multi-instance allocation of a distributed training job belongs to SageMaker Training, which Pipelines invokes as a step. Crediting the pipeline product with its downstream service's placement is the error this band avoids.

• https://docs.aws.amazon.com/sagemaker/latest/dg/pipelines.html recorded 2026-09-15

  AWS's Pipelines documentation describing the step types, the visual editor and SDK authoring, and the automatic provisioning of orchestration compute.