Overview
MLOps is a set of tools and methodologies for combining Machine Learning and DevOps practices. The goal of MLOps is to automate and optimize the deployment and maintenance of ML systems in production.
This instructor-led, live training (online or onsite) is aimed at engineers who wish to evaluate the approaches and tools available today to make an intelligent decision on the path forward in adopting MLOps within their organization.
By the end of this training, participants will be able to:
- Install and configure various MLOps frameworks and tools.
- Assemble the right kind of team with the right skills for constructing and supporting an MLOps system.
- Prepare, validate and version data for use by ML models.
- Understand the components of an ML Pipeline and the tools needed to build one.
- Experiment with different machine learning frameworks and servers for deploying to production.
- Operationalize the entire Machine Learning process so that it’s reproduceable and maintainable.
Format of the Course
- Interactive lecture and discussion.
- Lots of exercises and practice.
- Hands-on implementation in a live-lab environment.
Course Customization Options
- To request a customized training for this course, please contact us to arrange.
Requirements
- An understanding of the software development cycle
- Experience building or working with Machine Learning models
- Familiarity with Python programming
Audience
- ML engineers
- DevOps engineers
- Data engineers
- Infrastructure engineers
- Software developers
Course Outline
Introduction
- Machine Learning models vs traditional software
Overview of the DevOps Workflow
Overview of the Machine Learning Workflow
ML as Code Plus Data
Components of an ML System
Case Study: A Sales Forecasting Application
Accessing Data
Validating Data
Data Transformation
From Data Pipeline to ML Pipeline
Building the Data Model
Training the Model
Validating the Model
Reproducing Model Training
Deploying a Model
Serving a Trained Model to Production
Testing an ML System
Continuous Delivery Orchestration
Monitoring the Model
Data Versioning
Adapting, Scaling and Maintaining an MLOps Platform
Troubleshooting
Summary and Conclusion