Deploying Scalable Machine Learning for Data Science

Course details

Machine learning models often run in complex production environments that can adapt to the ebb and flow of big data. The tools and practices that help data scientists rapidly build machine learning models are not sufficient to deploy those models at scale. To deliver scalable solutions, you need a whole new toolset. This course provides data scientists and DevOps engineers with an overview of common design patterns for scalable machine learning architectures, as well as tools for deploying and maintaining machine learning models in production. Instructor Dan Sullivan reviews three technologies that enable scalable machine learning: services that expose models through APIs, containers for deploying models, and orchestration tools like Kubernetes that help manage containers and clusters. Plus, get tips for monitoring the performance of your services in production environments.
Learning objectives
Defining scalability
Tools and techniques for scalable machine learning
Architecture design patterns for scalable systems
Machine learning models as services
Containerizing models
Kubernetes for container orchestration
Monitoring performance
Best practices for scaling machine learning models

Concepts

0. Introduction

01 - Scaling ML models
02 - What you should know

1. The Need to Scale ML Models

03 - Building and running ML models for data scientists
04 - Building and deploying ML models for production use
05 - Definition of scaling ML for production
06 - Overview of tools and techniques for scalable ML

2. Design Patterns for Scalable ML Applications

07 - Horizontal vs. vertical scaling
08 - Running models as services
09 - APIs for ML model services
10 - Load balancing and clusters of servers
11 - Scaling horizontally with containers

3. Deploying ML Models as Services

12 - Services encapsulate ML models
13 - Using Plumber to create APIs for R programs
14 - Using Flask to create APIs for Python programs
15 - Best practices for API design for ML services

4. Running ML Services in Containers

16 - Containers bundle ML model components
17 - Introduction to Docker
18 - Building Docker images with Dockerfiles
19 - Example Docker build process
20 - Using Docker registries to manage images

5. Scaling ML Services with Kubernetes

21 - Running services in clusters
22 - Introduction to Kubernetes
23 - Creating a Kubernetes cluster
24 - Deploying containers in a Kubernetes cluster
25 - Scaling up a Kubernetes cluster
26 - Autoscaling a Kubernetes cluster

6. ML Services in Production

27 - Monitoring service performance
28 - Service performance data
29 - Docker container monitoring
30 - Kubernetes monitoring

Conclusion

31 - Best practices for scaling ML
32 - Next steps