MLOps Coffee Sessions #2: Different Ways of Serving ML Models with Byron Allen

In this session, Demetrios, David and Byron sat down to go over the 3 different ways of serving Machine Learning models as proposed in this blog post.

http://bugra.github.io/posts/2020/5/2...

Show notes

Microservices Guide - https://martinfowler.com/microservices/
Byron's Medium -   / byron.allen  

-------------- ✌️Connect With Us ✌️ ------------
Join our slack community: https://go.mlops.community/slack
Follow us on Twitter: @mlopscommunity
Sign up for the next meetup: https://go.mlops.community/register
Catch all episodes, Feature Store, Machine Learning Monitoring and Blogs: https://mlops.community/

Connect with Demetrios on LinkedIn:   / dpbrinkm  
Connect with David Aponte on Linkedin:
  / aponteanalytics  
Connect with Byron on Linkedin:
  / byronaallen  

Timestamps:
[00:00] Recap
[01:01] Takeaways from the blogpost of Bugra Akyildiz
[06:11] Brief introduction for Demetrios, Byron, and David
[08:16] Materializing or computing predictions offline
[11:40] Example of the first approach
[14:07] Applications of Serving ML Models
[16:15] "ML Introduction"
[17:25] Advantages and disadvantages
[20:46] Standout facts from the first approach
[24:01] Embedded model
[25:56] Security feature
[31:30] Serving Models
[33:50] "Sidecar"
[37:00] Microservice model serving
[46:20] Advantages of the third approach
[48:38] What's a container?
[51:21] The difference in the ability to scale
[1:00:10] Slack community promotion
[1:02:31] Wrap up