Amazon SageMaker Adds AI Data Preparation Capabilities via Python SDK APIs
Amazon SageMaker now includes general availability support for AI-driven data labeling, synthetic data generation, dataset quality assessment, and transformation through Python SDK APIs, enabling improved dataset workflows for model development.
What changed?
Amazon SageMaker now offers AI data preparation capabilities accessible from Python SDK APIs and an agent skill. Users can automatically label data, invite human review for low-confidence cases, generate synthetic examples, assess and improve existing datasets, and transform data at scale from their own notebooks or IDEs. These capabilities are available in the US East (N. Virginia) and US West (Oregon) AWS regions.

Why does it matter to an everyday developer?
Preparing high-quality training and evaluation data is often time-consuming and error-prone. The new features help developers automate many manual steps, reduce the need for custom scripts, and improve data quality. Developers can chain together synthetic data generation, automated and expert labeling, data quality checks, and transformations directly in their Python workflows. This integration supports faster iteration and higher trust in datasets for machine learning projects.
What can the developer do now?
- Use the SageMaker Python SDK to access AI data preparation capabilities across labeling, transformation, quality assessment, and synthetic data generation.
- Automate labeling at scale with supported and custom task types, and route uncertain cases to human review.
- Generate task-specific synthetic datasets from specifications and reuse data generation pipelines.
- Evaluate dataset quality with built-in and customizable metrics for duplicates, noise, and leakage.
