1.
MLOps with GitHub Actions
and Azure Machine Learning
Azure ML + GitHub Actions
Pulkit Agarwal, Product Manager @ GitHub

2.
MLOps ?

3.
DevOps for ML

4.
What is DevOps?
DevOps is the union of
people, processes, and
technologies
to deliver continuous
value to users
Development + operations

5.
What is MLOps?
Union of people, processes, and technologies
to deliver continuous value to users
using a ML based software application

6.
Is MLOps challenging ?
Why ?

7.
Building a model

8.
Building
a model
Data ingestion Data analysis
Data
transformation
Data validation Data splitting
Model
validation
Training
at scale
Logging
Deploying the
model
Monitoring
Machine Learning Applications Lifecycle

9.
Cat or Not ?

10.
Collaboration
on code
Remote
Training
Models
bookkeeping
(Many!)
Manual Steps
Managing
Data and Code
Updates
Challenges with MLOps

11.
MLOps sounds hard ?

12.
Lets make MLOps
Easy-ish

13.
Lets make MLOps
Easy-ish

14.
Collaboration
on code
Remote
Training
Models Book
keeping
(Many!)
Manual Steps
Data and Code
Updates
Making MLOps Easy(-ish)

15.
Making MLOps Easy(-ish)
+ +
ML
Optimized
Compute
Source
Control
ML
Aware
CI/CD

16.
Making MLOps Easy(-ish)
+ +
Source
Control
ML
Aware
CI/CD

17.
Making MLOps Easy(-ish)
+ +
ML
Aware
CI/CD

18.
Making MLOps Easy(-ish)
+ +

19.
DEMO
aka.ms/aidevday-mlops

20.
Thank You
Happy to take any questions !
GitHub Handle: @pulkitaggarwl || Email ID: puagarw@microsoft.com