Skip to main content

AWS Step Functions

AWS Step Functions is a visual workflow service that makes it easy to orchestrate over 220 AWS services and HTTPS endpoints such as SaaS applications into scalable, reliable, and resilient application. It supports common architectural and workflow patterns which makes it easy to coordinate the components of distributed applications as a series of durable steps in a visual workflow. Step Functions' workflows are written using Amazon States Language (ASL), defined as state machines, composed of steps called state, and can be used to orchestrate multiple AWS services.

Step Functions gives developers the ability to build and update applications quickly by managing the logic and implementing branching, parallel execution, and timeouts. Step Functions can also manage state, checkpoints, and restarts for you to make sure your application executes in order and as expected. It has built-in try/catch, retry, and rollback capabilities to help you deal with errors and exceptions automatically.

If you have a workload that requires co-ordinating distinct tasks, aggregation of results, fan-in and fan-out patterns, or that require human intervention, you may consider using Step Functions.

Common use cases include large-scale data processing, orchestration of microservices to build event-driven architectures, create data and machine learning pipelines, integration with SaaS applications, build generative AI applications, and automate IT security and processes.

No matter whether you are new to Step Functions or you already have a use case in mind, choose your own path and follow the curated learning steps to get started on Step Functions for a few of the common use cases.

Page topics

Path 1: Data Processing

Open all

Watch this tutorial to learn the basics on how to achieve data processing using Distributed Map for Step Functions. Understand the benefits of distributed data processing with serverless, learn different patterns, explore use cases, and discover performance optimization techniques.

Explore a sample project to learn about using Distributed Map for orchestrating large-scale parallel workloads, or use it as a starting point for your own projects. Distributed Map state can iterate over 10,00 rows of a CSV file that is generated using a Lambda function.

Take the Large-scale Data Processing with Step Functions workshop to learn about how to build large-scale data processing solutions with serverless such as AWS Step Functions, AWS Lambda, and AWS Fargate.

Path 2: Event-Driven Architecture

Open all

Watch this video to learn how to build real-life asynchronous architectures. Explore how choreography can help and how to handle transactions and workflows into your architectures with orchestration. See how both of these approaches work together.

Insurance Claims Processing Blogs
Blog 1
Blog 2
Blog 3

Use Step Functions to send a custom event to an event bus that matches a rule with multiple targets (Amazon EventBridge, AWS Lambda, Amazon Simple Notification Service, Amazon Simple Queue Service) to help build event-driven architectures.

Learn how to use EventBridge Pipes to launch a workflow with a message coming from SQS Queue.

Learn how to use EventBridge Rules to launch a workflow to process an object uploaded to S3

In this workshop, you will deploy a serverless backend that supports a pop-up coffee shop.You will use AWS Step Functions Workflow Studio to visually build the workflow that manages the drink orders through production. You will also learn how to emit events to a serverless event bus using AWS Step Functions.

Path 3: Data and ML Pipelines

Open all

In this project, learn how to use SageMaker and Step Functions to train a machine learning model and how to batch transform a test dataset.

In this project, learn how to use SageMaker to tune the hyperparameters of a machine learning model, and to batch transform a test dataset.

Learn how to automate and deploy custom ML models using service integrations between AWS Services using a CI/CD pipeline.

Learn how to detect online transaction fraud with serverless and machine learning technologies.

Use a Step Functions workflow to create a dataset and then train, evaluate, and use a Rekognition Custom Labels model. The workflow allows application developers and ML engineers to automate the custom label classification steps for any computer vision use case.

Take a workshop to learn how to use machine learning to automate and process documents at scale.

Path 4: Build generative AI applications

Open all

Learn how to use Step Functions to orchestrate recording file processing to generate insights using Amazon Bedrock or the LLM of your choice.

Build a Step Functions workflow for intelligent data processing using Textract and Foundational models on SageMaker

Leverage this sample project to demonstrate how you can integrate with Amazon Bedrock to perform AI prompt-chaining.

Read this blog to learn how you can leverage Step Functions workflows to incorporate human-in-the-loop processes to incorporate human judgment into your generative AI applications.

Take this workshop to gain a hands-on experience with building production-ready generative AI applications.

Path 5: Step Functions 101

Open all

Take the Step Functions Workshop to learn how to use the primary features of Step Functions through a series of interactive modules.

Get hands on experience with Step Functions Workflow Studio, a low code visual designer for Workflows. In this demo, you will create, run, and inspect a Hello World workflow in under 3 minutes.

Learn how to effectively use JSONPath and data filtering in Step Functions.

Learn about architectural best practices and repeatable patterns for building workflows and cost optimizations.