LIVE

GS Caltex Cuts MLOps Deployment Times Using Amazon SageMaker

🌐 Work & Society 1 min read 👁 10

South Korean energy firm GS Caltex has streamlined its machine learning operations by connecting Amazon SageMaker Canvas and Pipelines. The change has reduced model deployment times from roughly a week to a few hours.

GS Caltex Cuts MLOps Deployment Times Using Amazon SageMaker🌐Work & Society

Photo: AWS Machine Learning Blog

Key points

  • GS Caltex bridges process engineers and ML engineers using Amazon SageMaker Canvas and Pipelines.
  • Deployment times dropped from one week to a few hours through automated handoffs.
  • Running batch inference every 30 minutes instead of continuous endpoints lowered costs.

South Korean energy and chemical producer GS Caltex has overhauled its machine learning operations (MLOps) by linking Amazon SageMaker Canvas and Amazon SageMaker Pipelines. The company splits machine learning tasks between process engineers, who construct and validate models visually without writing code, and ML engineers, who handle production workflows and deployments.

Previously, moving assets from SageMaker Canvas into production pipelines required extensive manual integration. GS Caltex automated this transition while reusing existing assets. ML engineers now take preprocessing flows and registered models, verify data formats, and approve them for deployment using Amazon Simple Storage Service and Amazon EventBridge for scheduling.

According to the company, this automation has shortened deployment windows from about a week down to a few hours. Additionally, GS Caltex executes batch inference every 30 minutes instead of maintaining continuously running endpoints, a change that has notably decreased inference expenses.

What to watch next: Look for further technical documentation or reference implementations on GitHub as industrial enterprises continue adopting no-code tools alongside automated MLOps pipelines.

Source AWS Machine Learning Blog — aws.amazon.com/blogs/machine-learning/production-mlops-at-gs-caltex-wit… ↗

Contact the newsroom

Got a tip, a proposal or a partnership question? Write to us and we will reply shortly.