MLOps Essentials: Model Deployment and Monitoring

Course details

Machine learning operations (MLOps) is one of the fastest growing subfields of artificial intelligence. As more and more models have been deployed in production, the need for a structured, agile, end-to-end, automated machine learning lifecycle has continued to grow. In this course, instructor Kumaran Ponnambalam shows you how to apply key concepts from MLOps to create structured, improved outcomes in your everyday workflow.

Explore the fundamentals of MLOps to get up and running on your next machine learning project. Find out why so many data scientists, engineers, and project managers are so excited about ML models, as you discover the ins and outs of successfully deploying and monitoring models on your own. From continuous delivery to model serving and continuous monitoring to drift management, Kumaran equips you with the skills you need to start practicing effective, fair, explainable, and responsible artificial intelligence.

Concepts

0. Introduction

01 - Getting started with MLOps
02 - Course coverage
03 - Review of MLOps lifecycle

1. Continuous Delivery

04 - An ML production setup
05 - Deployment pipelines
06 - Deployment rollout strategies
07 - Planning for infrastructure
08 - Deployment best practices
09 - Tools and technologies for deployment

2. Model Serving

10 - Model serving patterns
11 - Scaling model serving
12 - Building resiliency in serving
13 - Serving multiple models
14 - Tools and technologies for serving

3. Continuous Monitoring

15 - The monitoring pipeline
16 - Instrumentation for observability
17 - Metrics to monitor
18 - ML production data best practices
19 - Alerts and thresholds for ML
20 - Tools and technologies for monitoring

4. Drift Management

21 - Introduction to model drift
22 - Concept drift basics
23 - Managing concept drift
24 - Feature drift basics
25 - Managing feature drift

5. Responsible AI

26 - Elements of responsible AI
27 - Explainable AI
28 - Fairness in ML
29 - Security of ML assets
30 - Privacy in machine learning

Conclusion

31 - Continuing on with MLOps