fr
en
MLOps on AWS
An AI model only delivers lasting value when its operation is well managed. With MLOps, connect deployment, monitoring and maintenance to organise the transition to production. Strengthen your ability to coordinate data and operations teams around models monitored over time.
- Duration
- 3 days 21 hours
- Code
- AWS11FR Code
Presentation
AWS provides tools and services to help teams build, train and deploy machine learning models efficiently. The MLOps approach draws on DevOps methodologies, emphasising automation, collaboration and monitoring throughout the model lifecycle. AWS offers services such as Amazon SageMaker, AWS Step Functions and other tools to simplify and optimise MLOps workflows.
This intensive course provides the practical skills to master MLOps engineering practices and use the associated AWS services. You will explore key MLOps concepts, automation tools and model deployment and monitoring strategies in depth. Through hands-on workshops, you will learn to configure CI/CD pipelines for machine learning, monitor model drift in production and automate model retraining.
By the end of this 3-day programme, you will have expertise in applying MLOps practices on AWS. You will master the key concepts, best practices and essential tools for optimising the deployment and management of machine learning models within your organisation.
As an Amazon Web Services authorised premium training partner (ATP), Oo2 offers skills development and certification courses that meet the organisation’s rigorous quality standards.
Objectives
By the end of this MLOps on AWS course, you will be able to:
- master MLOps principles and benefits, and establish the key distinctions from DevOps;
- design and implement security and governance strategies tailored to machine learning projects on AWS;
- configure and optimise MLOps experimentation environments using Amazon SageMaker;
- apply best practices for version management and ensuring the integrity of essential ML model components (data, code and models);
- build and automate CI/CD pipelines for ML, including model packaging, testing and deployment;
- automate ML model testing, packaging, deployment, degradation detection and retraining, and implement model monitoring.
Program
Module 1: understand the fundamentals of MLOps engineering practices
- Processes:
- machine learning workflows;
- automating recurring tasks;
- continuous integration and continuous delivery (CI/CD) for machine learning;
- model lifecycle management.
- People:
- roles and responsibilities within an MLOps team (Data Scientists, ML Engineers, DevOps Engineers);
- collaboration and communication between teams;
- the importance of DevOps culture in the machine learning context.
- Technology:
- MLOps tools and platforms (AWS SageMaker, etc.);
- cloud infrastructure for machine learning;
- programming languages and frameworks (Python, TensorFlow, PyTorch).
- Security and governance:
- sensitive data management;
- access control and authentication;
- regulatory compliance;
- model auditability and traceability.
Lab:
- Provision a SageMaker Studio environment with AWS Service Catalog.
Module 2: set up MLOps experimentation environments in SageMaker Studio
- The contribution of MLOps to experimentation.
- Setting up the ML experimentation environment.
- Creating and updating a lifecycle configuration for SageMaker Studio (demo).
Lab:
- Provision the experimentation environment.
Module 3: organise repositories for repeatable MLOps
- Data management for MLOps.
- ML model version control.
- Code repositories in ML.
Module 4: orchestrate ML pipelines for repeatable MLOps
- ML pipelines.
- Using SageMaker Pipelines to orchestrate model-building pipelines (demo).
- End-to-end orchestration with AWS Step Functions.
- End-to-end orchestration with SageMaker Projects.
- Standardising an end-to-end ML pipeline with SageMaker Projects (demo).
- Using third-party tools for reproducibility.
- Exploring human intervention during inference (demo).
- Governance and security.
- Security best practices for SageMaker (demo).
- Implementing repeatable MLOps.
Lab:
- Automate a workflow with Step Functions.
Module 5: ensure MLOps reliability
- Scaling and multi-account strategies.
- Testing and traffic shifting.
- Using SageMaker Inference Recommender (demo).
- Multi-account strategies.
Labs:
- Test model variants.
- Shift traffic.
Module 6: ensure MLOps reliability (monitoring)
- The importance of monitoring in ML.
- Operational considerations for model monitoring.
- Resolving issues identified through ML solution monitoring.
- Best practices for reliable MLOps.
Labs:
- Monitor a model for data drift.
- Build and troubleshoot an ML pipeline.
Audience
This course is intended for:
- MLOps engineers who want to deploy ML models to production and monitor them in the AWS cloud;
- DevOps engineers responsible for successfully deploying and maintaining ML models in production.
Prerequisites
The following prerequisites are recommended for this AWS course:
- completion of the AWS Technical Essentials course ;
- completion of the DevOps Engineering on AWS course, or equivalent experience ;
- completion of the Practical Data Science with Amazon SageMaker course, or equivalent experience.
Teaching and assessment methods
- Initial skills assessment
- Training materials provided to participants
- Continuous assessment throughout the course
- End-of-course feedback questionnaire
- Combination of theory and practical application
- Attendance records
- Post-course follow-up evaluation
- Practical exercises
Course highlights
- Expert AWS-certified instructors: benefit from recognised AWS-certified instructors with in-depth knowledge of AWS services for implementing MLOps practices.
- Interactive hands-on learning: master MLOps tools and techniques on AWS through practical demonstrations and workshops. Prepare to tackle real-world challenges in building, training, deploying and monitoring ML models in an MLOps environment.
- Develop key skills: the course content is carefully designed to build essential skills in planning and designing MLOps pipelines, using Amazon SageMaker, automating with AWS Step Functions and monitoring ML models.
Dates and sessions
Choose the date and delivery format that suit you.
No upcoming sessions are currently available.
AWS, Amazon SageMaker, Amazon Step Functions and other AWS marks are registered trademarks of Amazon.com, Inc. or its affiliates.