Skip to content

SYSTEM Cited by 4 sources

AWS Step Functions

What it is

AWS Step Functions is AWS's managed serverless workflow / state-machine service. Workflows are defined declaratively in Amazon States Language (ASL); steps invoke Lambda, AWS service integrations, or other Step Functions; failure/retry/catch/parallel constructs are first-class. Billed per state-transition.

Role in account-per-tenant lifecycle

ProGlove uses Step Functions as the account-creation orchestrator:

"Account creation is a fully automated process using AWS Step Functions, but the retirement and closure of accounts are performed manually through regularly run scripts." (Source: sources/2026-02-25-aws-6000-accounts-three-people-one-platform)

The Step Functions state machine wraps the multi-step account-provisioning flow: create account via Organizations → wait for account activation → apply baseline CloudFormation stacks (networking, IAM roles, logging, tagging, etc.) → register with StackSets as a new target → seed tenant-specific data → notify platform tooling.

The choice of Step Functions for this specific workflow — and plain scripts for retirement — is the architectural signal: the criterion for adoption is overhead introduced, not dogma. Step Functions shines where there are async waits, per-step retries, and observable progress. Retirement is a batch cleanup that runs regularly and doesn't need the state-machine scaffolding.

Stub page — general Step Functions internals (ASL semantics, Express vs Standard workflow, service-integration patterns) out of scope.

Seen in

  • sources/2026-02-25-aws-6000-accounts-three-people-one-platform — canonical instantiation of patterns/automate-account-lifecycle: Step Functions drives account creation; scripts handle account retirement.
  • sources/2026-02-26-aws-santander-catalyst-platform-engineering — the cloud process orchestration workload on Santander Catalyst migrates "legacy workflows to AWS Step Functions" with "retry patterns and error handling" + "centralized process monitoring". Step Functions is the modernization target for legacy batch / workflow engines in regulated-enterprise settings — its retry/catch/parallel primitives replace hand-rolled state machines.
  • sources/2026-04-01-aws-automate-safety-monitoring-with-computer-vision-and-generative-ai — two Step Functions workflows in the CV-safety pipeline: (1) GT Job creation triggered by EventBridge on configurable cadence, orchestrates SageMaker Ground Truth labeling jobs + post-processing Lambdas; (2) hourly tape-labeling preparation — identifies newly-onboarded cameras, analyses multiple time- shifted frames + object-detection predictions via voting mechanism, stitches clear regions into composite images showing unobstructed floor tapes for annotation.
  • — Zalando Payments 2021 deferred-payment risk-scoring pipeline uses Step Functions state machines as the orchestrator for five-stage ML workflows (Databricks preprocessing → scikit-learn batch transform → SageMaker training → SageMaker batch transform → SageMaker endpoint deployment). Authored via systems/zflow.
  • sources/2022-04-18-zalando-zalandos-machine-learning-platform — architectural decision statement on Zalando's choice of Step Functions for ML pipelines: "In early 2019 we at Zalando decided to use AWS Step Functions for orchestrating machine learning pipelines." Named fit reason: Zalando already used AWS as its main cloud provider, and Step Functions' service integrations covered the full ML workflow surface (Lambda, S3, SageMaker, Databricks via Lambda). Since early 2019, zflow has been used to compile hundreds of pipelines into Step Functions state machines. This is the 2022 platform-altitude disclosure of Step Functions as Zalando's ML pipeline runtime.
Last updated · 542 distilled / 1,571 read