SageMaker Feature Store adds batch writes and record listing
Amazon SageMaker Feature Store now supports batch writing records and listing existing records, streamlining high-throughput feature ingestion and discovery for AI models. Data scientists can ingest features at scale with fewer API calls and audit feature group contents more easily. These capabilities are available in all AWS Regions where SageMaker Feature Store is offered.
- →BatchWriteRecord for high-throughput feature ingestion
- →ListRecords for browsing and auditing feature group contents
- →Custom table and database naming for offline store
- →General availability in all SageMaker Feature Store regions
Features (3) ›
- BatchWriteRecord for high-throughput feature ingestion
Data scientists can now write multiple records across feature groups in a single request using BatchWriteRecord. This capability targets the online and/or offline stores, handles individual record failures, and supports TTL settings at multiple levels.
- ListRecords for browsing and auditing feature group contents
The new ListRecords API allows data scientists to retrieve record identifiers in a feature group paginated, enabling auditing, record lifecycle management, and recovery of identifiers without prior knowledge.
- Custom table and database naming for offline store
When configuring the offline store, users can now create Glue and Iceberg tables with custom names, offering more flexibility in managing data organization.
Notes (1) ›
- General availability in all SageMaker Feature Store regions
These new capabilities for ingestion and record management are available in all AWS Regions where Amazon SageMaker Feature Store is currently offered.
https://aws.amazon.com/about-aws/whats-new/2026/07/amzn-sgm-feature-store-batch-write-list/