Tutorial for async batched inference using sagemaker

[dwarez]

Note

Low Risk
Documentation-only addition (Jupyter notebook); no production code, auth, or library behavior changes.

Overview
Adds a new SageMaker SDK tutorial notebook under docs/sagemaker/notebooks/sagemaker-sdk/async-inference-embedding-tei/ that walks through batch corpus embedding for semantic search using asynchronous inference and Text Embeddings Inference (TEI).

The notebook explains async vs sync inference, then implements an end-to-end flow: load a slice of sentence-transformers/natural-questions, build and deploy BAAI/bge-small-en-v1.5 via ModelBuilder with AsyncInferenceConfig, configure scale-to-zero autoscaling (backlog target tracking plus wake-from-zero alarms), submit batched invoke_async jobs via S3, poll results, write a JSONL index to S3, and run a cosine-similarity retrieval smoke test. Optional SNS hooks and a CLEANUP-controlled teardown of endpoint, model, autoscaling, and CloudWatch alarms are included.

^{Reviewed by Cursor Bugbot for commit 151d62d. Bugbot is set up for automated code reviews on this repo. Configure here.}

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.