Learn MLOps in 4 Hours
Published by Pearson
Best Practices for Machine Learning and DevOps
In this training, we’ll cover everything from running MLOps experiments locally using remote data registries, to tracking experiment metrics in the cloud, and sharing your work with others. You’ll learn how to handle many of the challenges involved with deploying machine learning models to production with industry-standard tooling. Some of the challenges with developing machine learning models include being able to reproduce model results on production, collaborating with co-workers, and having issues with data quality.
By the time you finish this training, you’ll understand the concepts behind MLOps, how to build your own pipeline for any project, and why the ability to reproduce your training experiments is so important. Machine learning and data science are not going anywhere, so it’s important that you stay up-to-date and know what the best practices and tools are.
What you’ll learn and how you can apply it
By the end of the live online course, you’ll understand:
- How to use MLOps to improve machine learning models
- How to build a machine learning pipeline
- Why MLOps makes it easier to develop more accurate and reproducible models
And you’ll be able to:
- Visualize metrics and collaborate on machine learning projects
- Version and sharing data across different machines
- Implement tools to make deep learning projects more manageable
This live event is for you because...
- You are a machine learning engineer and you want to make your models better in production
- You are a data scientist who wants to version and share your data with other collaborators
- You are a DevOps engineer who wants to expand their skillset to handle machine learning projects
Prerequisites
- You have mid-level experience working on machine learning projects
- You know how to compare metrics for different training experiments
- You have some experience working in cloud environments (AWS, GCP, etc) and with GitHub
Course Set-up
- Make sure that you have the most recent version of Python 3 installed on your machine. You can download it here: https://www.python.org/downloads/
- You’ll also need to clone this repo to follow along with the class: https://github.com/flippedcoder/mlops-demo
Recommended Preparation
- Watch: Applied Continuous Delivery by Josh Long, Marcin Grzejszczak: https://www.oreilly.com/videos/applied-continuous-delivery/9780134857053/
- Read: Pragmatic AI: An Introduction to Cloud-Based Machine Learning, First Edition by Noah Gift: https://www.oreilly.com/library/view/pragmatic-ai-an/9780134863924/
- Watch: From IT to DevOps by Paul Furlan: https://learning.oreilly.com/videos/from-it-to/9780137606900/
Recommended Follow-up
- Read: Agile Application Lifecycle Management: Using DevOps to Drive Process Improvement by Bob Aiello, Leslie Sachs: https://www.oreilly.com/library/view/agile-application-lifecycle/9780132761857/
Schedule
The time frames are only estimates and may vary according to how the class is progressing.
Segment 1: Best practices and tools for machine learning projects (30 minutes)
- Setting up the project structure and the development environment
- Best ways to handle code and documentation
- How MLOps simplifies continuing machine learning model experiments
Q&A (5 min)
Break (5 min)
Segment 2: Pipeline automation and configuration management (50 minutes)
- Why you should automate your pipelines
- Example of an MLOps pipeline
- How to handle configurations for development
- Building a pipeline from scratch
- Running experiments in your pipeline
- Understanding metrics for your experiments
- Testing your pipeline
- Preventing unexpected behavior in your pipeline
Q&A (10 min)
Break (10 min)
Segment 3: Running experiments in the cloud (50 minutes)
- Making experiments reproducible with different tools
- Working with remote data registries
- Managing collaboration across teams
- Offloading experiments to the cloud
Q&A (10 min)
Break (10 min)
Segment 4: Working with deep learning projects in MLOps (50 minutes)
- Use cases
- Tracking metrics for training epochs
- Resuming training from a previous epoch
- Handling changes from the code and the data
Q&A (5 min)
Course wrap-up and next steps (5 minutes)
Your Instructor
Milecia McGregor
Milecia is a senior software engineer with a master’s in mechanical and aerospace engineering. She has worked in robotics, front-end development, back-end development, DevOps, IoT, machine learning, data science, cybersecurity, and almost every other part of tech. She’s also helped manage teams of developers and has done work as a developer advocate for a number of startups. In her free time, she likes to play with her dog and learn random skills like unicycle riding.