Practical Devops for the busy data scientist: Alice in DataOps land

Everyone uses the buzzword DevOps. Pretty much every company has hired or is about to hire a DevOps engineer or specialist. But how does this translates to Machine learning? What is more, how does this relate to your models' development, training, and deployment?

In this talk, we will debunk some of the DevOps terms and myths. The talk will include Machine learning examples and terms will be explained in a suitable way for data scientists/machine learners/ anyone willing to understand from scratch what is DevOps and how to use it.

Transcript

1. Tania Allard, PhD @ixek Developer Advocate @Microsoft Google Developer expert ML -Tensorflow Practical DevOps for the busy Data Scientist OSCON 2019 http://bit.ly/OSCON-MLOps
2. 2 Story time…. @ixek Down the rabbit hole
3. 3 @ixek
4. 4 A common story... @ixek Model / application to be productised R&D - develop, iterate fast, usually local or cloud Magic Is it live??
5. 5 @ixek
6. 6 Replacing the magic @ixek Model/app ready to productise R&D - develop, iterate fast, usually local or cloud MLOPs, automation, controlled deployment Worry free deployment! Wait and relax
7. 7 @ixek How skills are perceived
8. 8 @ixek
9. 9 @ixek
10. 10 DevOps is the union of people, process, and products to enable continuous delivery of value into production What is DevOps anyway? @ixek
11. 11 Sort of DevOps applied to data-intensive applications. Requires close collaboration between engineers, data scientists, architects, data engineers and Ops. How does it fit for DS? @ixek
12. 12 Story time…. The advice… getting started with MLOps
13. 13 @ixek MlOps Aims to reduce the end-to-end cycle time of data analytics/science from the origin of ideas to the creation of data artifacts.
14. 14 @ixek What to automate? Establish checkpoints Find the low hanging fruits How stable and robust are my processes? Devise a long term strategy What can I readily improve? Where am I? Can I count? Getting started
15. 15 It’s all madness @ixek
18. 18 @ixek Practical steps
19. 19 Keep everything in source control (data, code, infrastructure) - but allow for experimentation @ixek
20. 20 @ixek
21. 21 @ixek
22. 22 Standardize and define your environments in code (conda, pipfiles, Docker) @ixek
23. 23 Use canonical data sources - always know what data you are using (where it comes and goes) @ixek
24. 24 @ixek
25. 25 Automate wisely @ixek
26. 26 What and when to automate? @ixek • What should we automate? • Define success and failure metrics • Go from simple to complex tasks • Evaluate and monitor
27. 27 https://xkcd.com/1205/
28. 28 @ixek
29. 29 Use pipelines for repeatability and explainability @ixek
32. 32 Deploy portable models @ixek
33. 33 @ixek
34. 34 Test continuously and monitor production: push left @ixek
35. 35 @ixek
36. 36 Summary @ixek 1. DataOps help create value and improve end-to-end ML 2. Start by identifying the low-hanging fruits and defining automation success 3. Choose the right tooling and processes 4. Leverage people and processes 5. Implement wisely
37. 37 Thank you @ixek http://bit.ly/OSCON-MLOps