based on the documentation here – https://aws.amazon.com/blogs/big-data/build-multimodal-search-with-amazon-opensearch-service/, it states that open search can call amazon bedrock to generate embeddings. it uses special ml connector to ingest data.
I have set up an open search serverless collection via terraform , see sample code below . once the open search collection/cluster is created . how can i set up a pipeline , preferably in terraform, to ingest data from s3 and call bedrock service and push it to my open search database, if possible? or do i have to manually or programmatically read data from s3 , call bedrock via api and then do a push to open search .

provider "aws" {
  region = local.region
}

module "opensearch_collection_private" {
  source = "../../modules/collection"

  name        = "${local.name}-private"
  description = "Example private OpenSearch Serverless collection"
  type        = "SEARCH"

  create_access_policy  = true
  create_network_policy = true
  network_policy = {
    AllowFromPublic = false
    SourceVPCEs = [
      aws_opensearchserverless_vpc_endpoint.example.id
    ]
  }

  tags = local.tags
}