Practical DevOps for the busy data scientist

Slide 1

Slide 1 text

Practical DevOps for the busy data Scientist

Slide 2

Slide 2 text

bit.ly/PyConDE-mlops Slides

Slide 3

Slide 3 text

What you’ll learn 01 02 Why MLOps/ DevOps ? Who is responsible? 03 04 Getting started Getting from A to B

Slide 4

Slide 4 text

About Me

Slide 5

Slide 5 text

Software engineering Algorithm Data Answers @ixek bit.ly/PyConDE-mlops

Slide 6

Slide 6 text

Machine learning Answers Data Algorithm @ixek bit.ly/PyConDE-mlops

Slide 7

Slide 7 text

Machine learning Answers Data Model @ixek bit.ly/PyConDE-mlops @ixek bit.ly/PyConDE-mlops

Slide 8

Slide 8 text

Machine learning Answers Data Model Answers Predictions @ixek bit.ly/PyConDE-mlops

Slide 9

Slide 9 text

The data cycle Magic? R&D Generation @ixek bit.ly/PyConDE-mlops

Slide 10

Slide 10 text

Anyone? @ixek bit.ly/PyConDE-mlops

Slide 11

Slide 11 text

A common scenario @ixek bit.ly/PyConDE-mlops

Slide 12

Slide 12 text

@ixek bit.ly/PyConDE-mlops

Slide 13

Slide 13 text

If you had one wish? @ixek bit.ly/PyConDE-mlops

Slide 14

Slide 14 text

Replacing the magic ML Ops and robust pipelines R&D Generation @ixek bit.ly/PyConDE-mlops

Slide 15

Slide 15 text

How skills are perceived @ixek bit.ly/PyConDE-mlops

Slide 16

Slide 16 text

Better @ixek bit.ly/PyConDE-mlops

Slide 17

Slide 17 text

How they really are @ixek bit.ly/PyConDE-mlops

Slide 18

Slide 18 text

DevOps is the union of people, process, and products to enable continuous delivery of value into production - Donovan Brown What is devops @ixek bit.ly/PyConDE-mlops

Slide 19

Slide 19 text

MlOps Aims to reduce the end-to-end cycle time and friction of data analytics/science from the origin of ideas to the creation of data artifacts. What is devops @ixek bit.ly/PyConDE-mlops

Slide 20

Slide 20 text

But I do not work in a big company with many ML engineers @ixek bit.ly/PyConDE-mlops

Slide 21

Slide 21 text

Build your own MLOps Platform @ixek bit.ly/PyConDE-mlops

Slide 22

Slide 22 text

No content

Slide 23

Slide 23 text

No content

Slide 24

Slide 24 text

Practical steps @ixek bit.ly/PyConDE-mlops

Slide 25

Slide 25 text

We have the notebooks in source control @ixek bit.ly/PyConDE-mlops

Slide 26

Slide 26 text

Your saviour Source control ● Code and comments only (not Jupyter output) ● Plus every part of the pipeline ● And Infrastructure and dependencies ● And maybe a subset of data @ixek bit.ly/PyConDE-mlops

Slide 27

Slide 27 text

Everything should be in source control!! Except your training data which should be a known, shared data source Do not touch the raw data! Not even with a stick Your saviour @ixek bit.ly/PyConDE-mlops

Slide 28

Slide 28 text

Deterministic environments @ixek bit.ly/PyConDE-mlops

Slide 29

Slide 29 text

Whatever that environment is @ixek bit.ly/PyConDE-mlops

Slide 30

Slide 30 text

Your laptop is not a production environment… so ensure reproducibility @ixek bit.ly/PyConDE-mlops

Slide 31

Slide 31 text

@ixek bit.ly/PyConDE-mlops

Slide 32

Slide 32 text

Use pipelines for repeatability and reproducibility @ixek bit.ly/PyConDE-mlops

Slide 33

Slide 33 text

ml.azure.com

Slide 34

Slide 34 text

@ixek bit.ly/PyConDE-mlops

Slide 35

Slide 35 text

@ixek bit.ly/PyConDE-mlops

Slide 36

Slide 36 text

Automate wisely @ixek bit.ly/PyConDE-mlops

Slide 37

Slide 37 text

Adopt automation ● Orchestration for Continuous Integration and Continuous Delivery ● Gates, tasks, and processes for quality ● Integration with other services ● Triggers on code and non-code events @ixek bit.ly/PyConDE-mlops

Slide 38

Slide 38 text

Complete pipeline @ixek bit.ly/PyConDE-mlops

Slide 39

Slide 39 text

Kubeflow example https://www.kubeﬂow.org/docs/azure/azureendtoend/ @ixek bit.ly/PyConDE-mlops

Slide 40

Slide 40 text

Build pipeline- https://azure.microsoft.com/en-us/services/devops/https://azure.microsoft.com/e n-us/services/devops/

Slide 41

Slide 41 text

Code event trigger @ixek bit.ly/PyConDE-mlops

Slide 42

Slide 42 text

Release / deploy @ixek bit.ly/PyConDE-mlops

Slide 43

Slide 43 text

In brief Deterministic environments Use pipelines Continuous integration and delivery Source control (done right) Code, infrastructure, everything! Ensure production readiness For repeatable workﬂows Detect errors early and seamless deployments @ixek bit.ly/PyConDE-mlops

Slide 44

Slide 44 text

Want to learn more? ● ml.azure.com ● https://azure.microsoft.com/en-us/services/devops/ ● https://docs.microsoft.com/en-us/azure/machine-learning/ser vice/concept-ml-pipelines @ixek bit.ly/PyConDE-mlops

Slide 45

Slide 45 text

Come talk to us! @ ixek [email protected]