Overview

What is Amazon S3 Vectors?

Amazon S3 Vectors Amazon S3 Vectors
Amazon S3 Vectors — a purpose-built vector storage capability within S3 that enables similarity search over embeddings without a separate vector database. is a purpose-built vector storage capability within Amazon S3. It lets you store, index, and query high-dimensional vector embeddings without provisioning or managing a separate vector database.

Your entire vector search infrastructure becomes two Terraform resources:

resource "aws_s3vectors_vector_bucket" "this" {
  vector_bucket_name = "my-vectors"
}

resource "aws_s3vectors_index" "this" {
  vector_bucket_name = aws_s3vectors_vector_bucket.this.vector_bucket_name
  index_name         = "embeddings"
  dimension          = 1024
  distance_metric    = "cosine"
  data_type          = "float32"
}

That’s it. No clusters, no capacity planning, no connection pooling.

The demo application

This walkthrough documents a working application: an AWS announcements briefing system that:

1. Fetches AWS What’s New and AWS News Blog RSS feeds daily
2. Embeds each article into a 1024-dimensional vector via Titan Embeddings V2 Titan Embeddings V2
   Amazon Titan Embeddings V2 — the Bedrock foundation model used to generate 1024-dimensional text embeddings.
3. Stores vectors in S3 Vectors with metadata (title, URL, date, feed)
4. Serves a web UI where authenticated users ask questions
5. Searches vectors for relevant articles, then generates grounded answers via Claude

Architecture

What Terraform manages

Resource	Purpose	File
aws_s3vectors_vector_bucket	Vector storage container	s3_vectors.tf
aws_s3vectors_index	1024-dim cosine index	s3_vectors.tf
aws_s3_bucket	Source document storage	s3.tf
aws_lambda_function (×2)	Ingest + Query handlers	lambda_ingest.tf, lambda_query.tf
aws_apigatewayv2_api	HTTP API with JWT auth	api_gateway.tf
aws_cognito_user_pool	User authentication	cognito.tf
aws_amplify_app	SPA hosting	amplify.tf
aws_cloudwatch_event_rule	Daily ingest schedule	lambda_ingest.tf

What this walkthrough covers

Section	Content
S3 Vectors	Concepts, Terraform resources, API operations, IAM permissions
RAG Pipeline	Ingest flow, query flow, embedding strategy, similarity scoring
Deployment	Prerequisites, terraform apply, Cognito users, web UI
Troubleshooting	Bedrock errors, common issues, debugging

Note

This walkthrough documents

terraform-aws-s3-vectors-rag-demo

. Clone that repo to follow along with deployment steps.