Accelerate Your Data and AI Flow by Connecting Amazon SageMaker Unified Studio to Visual Studio Code
Accelerate Your Data and AI Flow by Connecting Amazon SageMaker Unified Studio to Visual Studio Code
by Lauren Mullennex, Anagha Barve, Anchit Gupta, and Bhargava Varadharajan — September 12, 2025 · Amazon SageMaker AI · Amazon SageMaker Unified Studio · Announcements · Intermediate (200) · Technical How-to
Developers and machine learning (ML) engineers can now connect their local Visual Studio Code (VS Code) editor directly to Amazon SageMaker Unified Studio. This capability lets you preserve your existing development workflow and personalized integrated development environment (IDE) configuration while still taking advantage of AWS analytics and AI/ML services in a single, unified data-and-AI studio. The integration provides seamless access from your laptop to scalable infrastructure for data processing, SQL analytics, and ML workloads. By linking your local IDE to SageMaker Unified Studio, you can optimize data and AI development pipelines without disrupting the best practices already in place.
In this post, we show how to connect your local copy of VS Code to SageMaker Unified Studio so you can build end-to-end data and AI workflows while continuing to work from your favorite IDE.
Solution overview
The architecture relies on three core components:
- Local workstation – Your development machine runs VS Code with the AWS Toolkit for Visual Studio Code and Microsoft Remote SSH installed. Use the Toolkit extension to browse SageMaker Unified Studio spaces and select the environment you want to work in.
- SageMaker Unified Studio – As part of the next generation of Amazon SageMaker, Unified Studio is a single experience where you can find and access your data, then work with familiar AWS tools for SQL analytics, data preparation, model development, and generative-AI application development.
- AWS Systems Manager – A secure, scalable remote-access and management service that links your local VS Code environment to SageMaker Unified Studio spaces and streamlines your data and AI workflows.
The diagram below shows how your local IDE interacts with SageMaker Unified Studio spaces.
Prerequisites
To try the remote IDE experience, make sure you have:
- Access to a SageMaker Unified Studio domain with internet connectivity. For domains configured in VPC-only mode, the domain must reach the internet through a proxy or NAT gateway. If your domain is completely isolated, follow the remote IDE support guide. You can create a domain via the quick setup or manual setup workflow.
- An SSO user authenticated through IAM Identity Center. See the user management documentation to configure SSO access.
- Access to, or the ability to create, a SageMaker Unified Studio project.
- A compute space (JupyterLab or Code Editor) with at least 8 GB of memory. In this walkthrough we use an
ml.t3.largeinstance. SageMaker Distribution image version 2.8 or later is supported. - The latest stable build of VS Code on your local machine with Microsoft Remote SSH (version 0.74.0 or later) and the AWS Toolkit extension (version 3.74.0).
Implement the solution
Complete the following steps to enable remote access and connect to a space from VS Code. A SageMaker Unified Studio space must have remote access turned on before you can reach it from your IDE.
- Open your JupyterLab or Code Editor space. If it is already running, stop the space and choose Configure space so you can enable remote access.
- Turn on Remote access and choose Save and restart.
- In your local VS Code, open the AWS Toolkit pane.
- On the SageMaker Unified Studio tab, choose Sign in, then supply your domain URL, for example
https://<domain-id>.sagemaker.<region>.on.aws.
- You are redirected to a browser to authorize the AWS IDE extensions. Choose Open to launch the new browser tab.
- Choose Allow access so the extensions can reach your project from VS Code.
- When you see Request approved, you have remote access to the domain.
Return to your local VS Code window to keep working on ETL jobs, data pipelines, ML training and deployment, or generative-AI applications. To connect to a project for data processing and ML development, perform these steps:
- Choose Select a project to view data assets and compute resources. All projects in the domain are listed, but you can only open the projects where you are a member.
You can view only one domain and one project at a time. To switch projects or sign out of the domain, select the ellipsis icon.
You can also review the data and compute resources you created earlier. - Connect to a JupyterLab or Code Editor space by choosing the connect icon. If the icon is hidden, remote access might be disabled for that space. If the space status is “Stopped,” hover over it and choose connect—this enables remote access, starts the space, and opens the connection. If the space is “Running,” stop it, enable remote access, and reconnect from the Toolkit.
VS Code opens a second window that connects to your SageMaker Unified Studio space over remote SSH. - Go to Explorer to see the notebooks, files, and scripts inside the space. From the AWS Toolkit you can also browse the data sources that belong to the project.
Use your custom VS Code setup with SageMaker Unified Studio resources
When VS Code connects to SageMaker Unified Studio, you keep all of your personal hotkeys, settings, and snippets. If you rely on snippets to insert common analytics or ML patterns, they continue to work while you run on SageMaker Unified Studio’s managed infrastructure.
In the following example we run analytics snippets: show-databases queries Amazon Athena to list available databases, show-glue-tables lists the tables stored in the AWS Glue Data Catalog, and query-ecommerce uses Spark SQL to fetch data for analysis.
You can also use snippets to automate ML model builds and training on SageMaker AI. The screenshot below shows snippets that handle data processing, configure training, and launch a SageMaker AI training job. This approach lets data practitioners keep their familiar dev setup while tapping into the managed data and AI resources inside SageMaker Unified Studio.
Disable remote access in SageMaker Unified Studio
If you are an administrator and need to disable this feature, add the following policy statement to the project’s IAM role:
{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "DenyStartSessionForSpaces",
"Effect": "Deny",
"Action": [
"sagemaker:StartSession"
],
"Resource": "arn:aws:sagemaker:*:*:space/*/*"
}
]
}
Clean up
By default, SageMaker Unified Studio stops idle resources such as JupyterLab and Code Editor spaces after 1 hour. If you created a dedicated SageMaker Unified Studio domain just for this walkthrough, remember to delete the domain when you finish.
Conclusion
Connecting your local IDE directly to Amazon SageMaker Unified Studio eliminates the friction of jumping between on-device development and scalable data-and-AI infrastructure. Because you keep your personalized IDE configuration, you no longer need to switch mindsets between environments. Whether you are processing large datasets, training foundation models, or building generative-AI applications, you can remain in your local setup while accessing all of SageMaker Unified Studio’s capabilities. Connect your IDE to SageMaker Unified Studio today to streamline data workflows and speed up ML development.
About the authors
