MLA-C01: AWS Machine Learning Engineer – Associate

Data ingestion and storage

• Training data formats and storage options
• S3, EFS and FSx for Lustre for training data
• Streaming data for ML
• Data validation basics

• Lab: Set up a SageMaker Studio domain and load a dataset · 35 min

Transformation and feature engineering

• Cleaning, encoding and scaling features
• SageMaker Data Wrangler and Processing jobs
• SageMaker Feature Store
• Handling imbalance and missing values

• Lab: Engineer features with Data Wrangler and publish to Feature Store · 55 min

Data quality and bias

• Pre-training bias metrics with Clarify
• Labelling with SageMaker Ground Truth
• Protecting sensitive data

Choosing an approach

• Built-in algorithms versus script mode
• Using foundation models with JumpStart and Amazon Bedrock
• AWS AI services as alternatives to custom models

Training and tuning

• Training jobs, instance types and Spot training
• Automatic model tuning
• Distributed training at a glance
• Experiments tracking

• Lab: Train and tune an XGBoost model with automatic model tuning · 60 min

Evaluation and explainability

• Classification and regression metrics
• Detecting overfitting and underfitting
• Explainability with SHAP and Clarify
• Model debugging

• Lab: Evaluate the model and generate an explainability report · 40 min

Deployment options

• Real-time, serverless, asynchronous and batch inference
• Multi-model and multi-container endpoints
• Auto scaling endpoints
• Shadow and canary deployments

• Lab: Deploy a real-time endpoint and a batch transform job · 45 min

Automation and monitoring

• SageMaker Pipelines
• Model Registry and approval workflows
• Model Monitor for data and quality drift
• CI/CD for ML with CodePipeline

• Lab: Build a training pipeline with a registry approval step · 55 min

Security, cost and exam readiness

• IAM roles, VPC isolation and encryption
• Cost controls for training and hosting
• MLA-C01 domain recap and practice set