About the course
Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to quickly build, train, and deploy machine learning models at scale. It is designed to cover the entire machine learning lifecycle, simplifying the process of bringing ML solutions from idea to production. This comprehensive 4-day training course provides a deep dive into the essential components of SageMaker, enabling participants to effectively leverage the platform on AWS for their machine learning projects.
You will gain extensive hands-on experience using key SageMaker tools and services, starting with navigating the environment in SageMaker Studio and efficiently preparing datasets using SageMaker Data Wrangler. The course provides in-depth coverage of model training, including using SageMaker's built-in algorithms and running custom code with popular frameworks like TensorFlow and PyTorch. You will learn crucial practices for debugging training jobs with SageMaker Debugger and optimising models through automatic hyperparameter tuning. Model validation is covered, including leveraging SageMaker Clarify for bias detection and explainability, alongside exploring automated machine learning with SageMaker Autopilot.
The workshop also delves into model deployment options, such as real-time endpoints and batch transforms, and introduces essential MLOps fundamentals using SageMaker Model Registry, Pipelines, and Projects for automating and managing ML workflows. Finally, you will explore SageMaker Feature Store for managing and discovering features. Through extensive hands-on labs throughout the four days, you will acquire the practical skills to confidently build, train, deploy, and manage machine learning models at scale using the Amazon SageMaker platform.
Instructor-led online and in-house face-to-face options are available - as part of a wider customised training programme, or as a standalone workshop, on-site at your offices or at one of many flexible meeting spaces in the UK and around the World.
-
- Navigate the SageMaker environment, including SageMaker Studio, and understand its role in the ML lifecycle.
- Prepare and analyse datasets for ML using SageMaker Data Wrangler.
- Train ML models using SageMaker's built-in algorithms and custom code with popular frameworks via SageMaker training jobs.
- Use SageMaker Debugger to monitor and debug training jobs.
- Configure and launch Automatic Model Tuning jobs to optimise models.
- Deploy models to real-time endpoints for online inference.
- Use SageMaker Batch Transform for offline inference.
- Perform model validation, including bias detection and explainability using SageMaker Clarify.
- Leverage SageMaker Autopilot for automated machine learning tasks.
- Understand MLOps concepts and use SageMaker Model Registry, Pipelines, and Projects to automate ML workflows.
- Manage and retrieve features using SageMaker Feature Store for training and inference.
- Understand key design considerations and common patterns for building ML solutions on AWS using SageMaker.
- Identify potential cost optimisation opportunities when using SageMaker services.
-
This comprehensive 4-day course is designed for individuals involved in building, training, and deploying machine learning models on AWS. It is ideal for:
Data Scientists looking to operationalise their models on AWS.
Machine Learning Engineers building and managing ML pipelines.
MLOps Engineers focused on automating the ML lifecycle.
Software Developers working with ML features in applications.
Solutions Architects designing ML systems on AWS.
Data Engineers supporting ML workloads on AWS.
-
Participants should have:
A foundational understanding of core Machine Learning and Deep Learning concepts.
Working knowledge of the Python programming language.
Basic familiarity with AWS services (e.g., S3, IAM, EC2 concepts) is beneficial but not strictly required, as relevant AWS concepts will be introduced in context.
Prior experience with ML frameworks like TensorFlow, PyTorch, or Scikit-learn is helpful but not mandatory.
No prior experience with Amazon SageMaker is required, as the course starts with introductory concepts.
-
This SageMaker course is available for private / custom delivery for your team - as an in-house face-to-face workshop at your location of choice, or as online instructor-led training via MS Teams (or your own preferred platform).
Get in touch to find out how we can deliver tailored training which focuses on your project requirements and learning goals.
-
Introduction to Machine Learning on AWS and SageMaker
Overview of key Machine Learning and Deep Learning concepts and the typical ML lifecycle.
Introduction to Machine Learning services available on AWS (overview of SageMaker within the broader AWS AI/ML stack).
What is Amazon SageMaker? Its purpose and how it supports the entire ML lifecycle.
Navigating the AWS Console and accessing SageMaker.
Exploring SageMaker Studio: The unified, web-based integrated development environment (IDE) for ML (Notebooks, Experiments, Debugger integration, etc.).
Understanding AWS S3: The primary data storage service for SageMaker – buckets, objects, data organisation.
Setting up necessary AWS Identity and Access Management (IAM) permissions for SageMaker users.
Hands-On Lab: Navigating the AWS Console and SageMaker Studio, setting up IAM roles/policies for SageMaker access, working with S3 buckets for data storage.
Data Preparation and Feature Engineering with SageMaker
Identifying your data sources and articulating your ML problem statement.
Understanding data formats and schemas suitable for SageMaker.
Uploading and managing datasets in S3 for efficient SageMaker access.
Introduction to SageMaker Data Wrangler:
Connecting to various data sources (S3, Redshift, etc.).
Importing and sampling large datasets.
Analysing data quality and understanding data types within Data Wrangler.
Performing data transformations and cleaning (handling missing values, outliers, encoding categorical data).
Feature Engineering: Creating new features using Data Wrangler's built-in tools.
Generating data quality and insights reports.
Exporting prepared data for training.
Hands-On Lab: Ingesting data into Data Wrangler from S3, performing common data transformations and cleaning steps, generating data analysis reports, exporting the processed dataset.
Model Training in SageMaker (Built-in Algorithms and Script Mode)
Overview of SageMaker Training Jobs: How training jobs work, inputs, outputs.
Assigning Compute Resources: Choosing appropriate instance types for training, understanding cost implications.
Introduction to Distributed Training concepts in SageMaker.
Using SageMaker's Built-in Algorithms: Overview of available algorithms and how to select one for common tasks.
Training with Custom Code using SageMaker Script Mode:
Preparing training scripts using popular frameworks like Python (Scikit-learn), TensorFlow, and PyTorch.
Using the SageMaker SDK Estimators to define and launch training jobs for custom code.
Understanding SageMaker Experiments: Tracking, comparing, and managing your training runs.
Hands-On Lab: Launching training jobs using a SageMaker built-in algorithm, preparing a custom training script and launching a training job using the SageMaker SDK Estimator, tracking training runs in SageMaker Experiments.
Model Debugging and Tuning
SageMaker Debugger:
Monitoring training jobs for common issues (e.g., overfitting, underfitting, gradient problems) using Debugger rules.
Configuring Debugger hooks and rules (built-in and custom).
Analysing Debugger outputs.
Model Tuning:
Defining appropriate metrics for model evaluation and tuning.
Hyperparameter Tuning Concepts.
Automatic Model Tuning (AMT) in SageMaker: Configuring and launching hyperparameter tuning jobs.
Analyzing tuning job results to identify the best performing model.
Hands-On Lab: Configuring and using SageMaker Debugger with a training job, launching and analysing a SageMaker Automatic Model Tuning job to find the best hyperparameters.
Model Deployment and Inference
Overview of SageMaker Deployment Options: Choosing the right method for your use case.
SageMaker Hosting Services (Real-time Endpoints):
Understanding SageMaker Models and Endpoint Configurations.
Configuring and creating HTTPS Endpoints for low-latency, real-time inference.
Selecting appropriate instance types for inference endpoints.
Testing deployed real-time endpoints.
Scaling Endpoints (Auto Scaling basics).
(Optional: Brief mention of Multi-Model Endpoints, Inference Pipelines).
SageMaker Batch Transform:
Using Batch Transform for offline inference on large datasets.
Configuring and launching Batch Transform jobs.
Making inferences from your dataset using both real-time endpoints and batch transform.
Hands-On Lab: Deploying a trained model to a real-time endpoint, testing the endpoint, configuring and running a Batch Transform job for offline inference.
Model Validation and Automated ML
Model Validation Strategy: Validating model performance using holdout sets in a deployment context.
Offline Testing vs. Online Testing / A/B Testing Concepts.
SageMaker Clarify:
Detecting potential bias in your data and trained models.
Gaining model explainability (SHAP, LIME).
Configuring and running Clarify jobs for bias detection and explainability reports.
Introduction to SageMaker Autopilot:
What is Automated Machine Learning (AutoML)?
Configuring and launching an Autopilot experiment.
Reviewing Autopilot results, understanding the generated models, and deploying the best performing one.
Hands-On Lab: Using SageMaker Clarify to analyse a dataset or model for bias/explainability, running a SageMaker Autopilot experiment and evaluating its output.
MLOps Fundamentals with SageMaker
Introduction to MLOps: Concepts, benefits, and challenges of taking ML models to production and maintaining them.
SageMaker Model Registry: Managing model versions, metadata, and approval status.
Introduction to SageMaker Pipelines:
Defining and automating multi-step ML workflows (data processing -> training -> tune -> evaluate -> model registration -> deployment steps).
Executing Pipeline runs and tracking lineage.
Introduction to SageMaker Projects: Using MLOps project templates for setting up CI/CD pipelines for ML using AWS CodeCommit, CodeBuild, and CodePipeline.
Hands-On Lab: Working with SageMaker Model Registry, creating a simple SageMaker Pipeline definition and executing a run, exploring SageMaker Projects templates.
SageMaker Feature Store and Course Wrap-up
SageMaker Feature Store:
Concepts: Feature Groups, Features, Online Store (low-latency inference), Offline Store (training, batch inference).
Creating Feature Groups.
Ingesting feature data (batch and real-time ingestion methods).
Retrieving features for training (Offline Store) and inference (Online Store).
Feature Discovery within Feature Store.
Review of the SageMaker Ecosystem and its role across the ML lifecycle.
Troubleshooting Common Issues across SageMaker services.
Cost Optimization considerations (e.g., Managed Spot Training).
Next Steps: Continuing your MLOps journey, exploring other AWS AI/ML services.
Hands-On Lab: Working with SageMaker Feature Store (creating feature group, ingesting sample data, retrieving data for training/inference).
-
Official Amazon SageMaker Documentation - https://docs.aws.amazon.com/sagemaker/
AWS Machine Learning Documentation - https://aws.amazon.com/documentation/machine-learning/
Amazon SageMaker Studio Documentation - https://docs.aws.amazon.com/sagemaker/latest/dg/studio.html
Amazon SageMaker Data Wrangler Documentation - https://docs.aws.amazon.com/sagemaker/latest/dg/data-wrangler.html
Amazon SageMaker Autopilot Documentation - https://docs.aws.amazon.com/sagemaker/latest/dg/autopilot.html
Amazon SageMaker Pipelines Documentation - https://docs.aws.amazon.com/sagemaker/latest/dg/pipelines.html
Amazon S3 Documentation - https://docs.aws.amazon.com/s3/
AWS IAM Documentation - https://docs.aws.amazon.com/iam/
Trusted by
