Introduction to Positron on Amazon SageMaker AI
Data science teams operating within enterprise cloud environments often require flexible development interfaces that handle both Python and R without sacrificing infrastructure control or security. Addressing this requirement, Positron now runs on Amazon SageMaker AI as a custom Docker image currently available in public preview. This integration brings Posit’s modern data science editor directly into governed cloud environments, allowing practitioners to query cloud data lakes, validate features, train machine learning classifiers, and document analytical workflows within a single cohesive workspace. By bridging the gap between specialized analytical tooling and managed cloud infrastructure, the integration supports diverse data science workloads while adhering to enterprise governance standards.
Product Definition and Architectural Foundation
Positron serves as Posit’s next-generation Python and R integrated development environment built specifically for data science practitioners. The environment operates inside Amazon SageMaker AI as a custom Docker image in public preview, constructed upon the underlying Amazon SageMaker Distribution image. This technical integration stems directly from close collaboration between Posit and the Amazon SageMaker team on running Positron inside Amazon SageMaker, supported by a Strategic Collaboration Agreement between Posit and AWS signed in September 2025. By combining Posit’s editor architecture with robust SageMaker infrastructure, data science teams gain a multilingual workspace that executes directly on dedicated cloud compute instances housed within their own Amazon Web Services accounts. Furthermore, this architectural alignment ensures that teams can leverage familiar development paradigms while maintaining strict oversight of compute resources and data access channels.
Preloaded Tooling and Licensing Structure
To support complex analytical tasks immediately upon deployment, the custom Docker image arrives preloaded with an extensive software stack. R is included natively alongside common packages sourced from Posit Package Manager to streamline library distribution. For Python environment management, the image incorporates uv for fast package installation and dependency resolution. In addition, the preloaded software stack includes Quarto for technical publishing, professional ODBC drivers for database connectivity, and a Bedrock-backed Posit Assistant to aid coding tasks. Because Positron bundled within this Docker image operates as a commercial product, it requires a Posit Workbench Advanced license in order to run, with licensing managed centrally through AWS License Manager. Responsibilities for maintaining the environment are clearly divided: Posit provides the software image and direct product support, while AWS operates the underlying managed cloud services, and the customer manages identity, permissions, licensing, networking, logging, image updates, and approved AWS services.
Environment Deployment, Availability, and Execution Roles
Deploying and configuring the environment requires adherence to specific infrastructure parameters and availability guidelines. The Positron Docker image runs today in Amazon SageMaker Studio and will be coming soon to Amazon SageMaker Unified Studio. When deploying the environment, practitioners must ensure their underlying infrastructure meets minimum prerequisites, such as utilizing an ml.t3.xlarge instance or larger for the demonstrated environment to ensure adequate compute capacity. Once properly configured, Positron runs under the designated Space execution role within a governed SageMaker Studio Space. Moreover, users can run multiple Spaces at once for independent projects or utilize a shared Space so several team members can collaborate simultaneously within the same Positron application instance.
Walkthrough Workflow and Data Integration
To understand how these architectural components interact in a production setting, consider a standardized data science walkthrough workflow. Users can query Amazon Athena, the AWS Glue Data Catalog, and Amazon S3 directly from the Positron IDE using Posit’s professional ODBC drivers, which are installed and available for supported connections to a variety of database platforms including Athena and Redshift. The walkthrough workflow utilizes a synthetic 50,000-loan portfolio stored in Amazon S3 and registered in the AWS Glue Data Catalog. Within this pipeline, Amazon Athena queries the raw data, R validates features, Python trains an XGBoost classifier, Shiny for Python invokes the resulting model endpoint, and Quarto records the complete workflow for reproducible reporting. Throughout these coding tasks, Posit Assistant acts as Posit’s AI coding assistant that can utilize Amazon Bedrock as its underlying model provider.
Model Training Results and Session Performance Metrics
Executing the loan portfolio walkthrough pipeline yields specific analytical and operational metrics that demonstrate the practical utility of the platform. Data profiling results identified 50,000 total loans, including 1,500 records with missing income data and 1,015 actual defaults, resulting in an overall baseline default rate of 2.03 percent. This left exactly 48,500 clean loans available for modeling and scoring after excluding the incomplete records. Following rigorous data preparation, an XGBoost classifier was trained on a feature matrix containing 40,000 rows and three primary model features. The subsequent held-out evaluation produced an Area Under the Curve of 0.834 and demonstrated a 12.3 percent observed default rate in the highest-risk decile.
Operational efficiency and resource utilization are tracked via the integrated Session information view within the editor. According to the captured Session information view, the active session recorded a total of 6,657,942 tokens, which included 6,118,411 cache-read tokens and 462,905 cache-write tokens. This execution incurred an estimated cost of $6.319 while achieving a cache efficiency rating of 92.5 percent, illustrating the cost-effective nature of running model development workflows inside the managed cloud environment.
Conclusion
The integration of Positron on Amazon SageMaker AI successfully brings Posit’s specialized IDE for data science together with Amazon SageMaker AI. By enabling data scientists to explore Amazon Athena tables, validate features using R, train XGBoost models in Python, deploy real-time SageMaker AI endpoints, and report findings with Quarto inside a governed SageMaker Studio Space, the solution provides a comprehensive platform for enterprise analytics. Through structured licensing, preloaded tooling, and robust cloud data integration, teams can execute complex machine learning projects efficiently while maintaining full administrative control over their cloud infrastructure.
Sources
- Run Positron on Amazon SageMaker AI for data science workflows
- Introducing Positron on Amazon SageMaker
- Run Positron on Amazon SageMaker AI for data science workflows | Amazon Web Services
Explore more Infrastructure coverage.
