Skip to content

S3 Vectors

The S3 Vectors storage provider uses AWS S3 Vector Buckets to store and query vector indexes.

Requirements

To use the S3 Vectors storage provider you must:

  • Have an AWS account with an S3 Vector Bucket configured
  • Have the boto3 Python package installed
  • AWS access credentials in your application environment

Usage

Instantiate a S3 Vectors provider with:

from django_ai_core.contrib.index.storage.s3vectors import S3VectorProvider

storage_provider = S3VectorProvider(
    bucket_name="my-bucket-name", dimensions=1536
)

Where bucket_name is the name of your pre-existing S3 Vector Bucket and dimensions is the number of dimensions output by your selected embedding model.

If using this storage provider outside of the context of a VectorIndex, you will also need to specify an index_name.