Build, train, and deploy ML models using Amazon SageMaker
summary
Overview of concrete SageMaker capabilities, not theory: Ground Truth auto labeling cuts labeling cost up to 70% and supports 30+ workflows; Data Wrangler exports transforms to code and addresses the ~80% time customers spend on data prep; training runs use transient compute with Managed Spot Training saves up to 90% and distributed training, plus Experiments and Debugger to track and stop poor runs. Deployment supports persistent endpoints, batch transform, Elastic Inference, A/B splits (example 66/33), multi-model endpoints, Model Monitor data capture and baselining, and Pipelines with model registry and CI/CD project templates.
AI-written summary · about
chapters (16)
- 0:00Intro
- 1:09Amazon SageMaker overview
- 2:37Model building and training
- 2:52Amazon SageMaker Ground Truth
- 4:5580% of time spent on data preparation
- 5:47Amazon SageMaker Data Wrangler
- 8:15Amazon SageMaker Processing
- 10:21Amazon SageMaker Feature Store
- 12:50SageMaker Studio notebooks
- 14:56Training algorithms and frameworks
- 17:01Amazon SageMaker Autopilot
- 22:00SageMaker managed deployments
- 27:18SageMaker multi-model endpoints
- 28:15Amazon SageMaker Model Monitor
- 30:34Amazon SageMaker Pipelines
- 32:40Learn ML with AWS Training and Certification
related talks
31mBuild, train, and deploy ML models using Amazon SageMaker
34mBuild, train and deploy Machine Learning models with Amazon SageMaker
32mBuild, Train, and Deploy Machine Learning Models on AWS
37mStreamlining Machine Learning Operations
45mDeploying ML Models to Production with Amazon SageMaker
39mDevelop your ML project with Amazon SageMaker
19mUsing Amazon SageMaker to develop and deploy ML models
31mTrain and tune ML models to the highest accuracy using Amazon SageMaker