[MLS-C01] [IMPL and OPs] Deploy and Operationalize

Deploy and operationalize machine learning solutions with lab

Posted by Oscaner on August 8, 2022


  1. Create your data
    • Gather, engineer, transform
  2. Produce your model
    • Build model and evaluate it
  3. Execute (operationalize) your model
    • Deploy, monitor, retrain

Deployment and Operationalization

  • Deploy
  • Inference
  • Metrics
  • Monitor performance
  • Analysis
  • Retrain



Model Monitor

Monitor your production model to detect deviations in data quality compared to a baseline dataset

  • Monitor your endpoint using Model Monitor (csv or flat-json)
    1. Baseline
    2. SageMaker suggests baseline constraints
    3. Create baselining job
    4. Create a continuous monitoring schedule
    5. Start continuous monitoring
  • Analyze / Retrain


  • To optimize the performance of your models you can use Automatic Model Tuning to find the hyperparameters that will give you the optimal performance
  • You can use SageMaker hosting to automatically scale your model to the performance needed for your model

本文由 Oscaner 创作, 采用 知识共享署名4.0 国际许可协议进行许可
本站文章除注明转载/出处外, 均为本站原创或翻译, 转载前请务必署名