← Back
AWS ML Blog

Announcing instance preference lists for Amazon SageMaker AI training jobs

•11 min read•
#amazon
Level:Intermediate
For:ML Engineers, DevOps
✦TL;DR

Amazon SageMaker AI introduces instance preference lists for training and processing jobs, letting users specify an ordered list of up to five instance types. When a job is submitted, SageMaker automatically launches it on the first instance type in the list that has available capacity, removing the need for manual retry loops or capacity‑watching scripts. This feature streamlines job scheduling and improves resource utilization across the SageMaker platform. The change is especially useful for teams that routinely run large-scale training workloads and need predictable launch times without custom orchestration code.

⚡ Key Takeaways

  • Users can list up to five instance types in a priority order for a training job.
  • SageMaker evaluates the list and selects the first instance type with free capacity.
  • The automatic selection eliminates manual retry loops that were previously required.
  • Capacity‑watching scripts are no longer necessary, simplifying job submission pipelines.
  • The feature applies to both training and processing jobs within SageMaker AI.
  • WhyItMatters: By automating instance selection, engineers can reduce operational overhead, achieve faster job start times, and improve overall resource efficiency in production training pipelines.
  • TechnicalLevel: Intermediate
  • TargetAudience: ML Engineers, DevOps
  • PracticalSteps:
  • Include an `instance_preference` list in the training job configuration JSON.
  • Submit the job via the SageMaker Python SDK or AWS CLI, letting Sage
💡 Why It Matters

By automating instance selection, engineers can reduce operational overhead, achieve faster job start times, and improve overall resource efficiency in production training pipelines.

✅ Practical Steps

  1. Include an `instance_preference` list in the training job configuration JSON.
  2. Submit the job via the SageMaker Python SDK or AWS CLI, letting Sage

Want the full story? Read the original article.

Read on AWS ML Blog ↗

More like this

Introducing managed cluster extensions for observability and user management in Amazon SageMaker HyperPod Slurm

AWS ML Blog•#deployment

Introducing Amazon SageMaker HyperPod Global Inference Gateway: One endpoint across Regions and clusters

AWS ML Blog•#deployment

Amazon SageMaker Inference: 2026 launches in review, part 2

AWS ML Blog•#deployment

Enable AI agents to pay for AWS Marketplace x402 products with Amazon Bedrock AgentCore payments

AWS ML Blog•#agents

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING