Employing Google Cloud Machine Learning Engine to Develop Models in Production

How Holiday Extras are utilising Google Cloud Machine Learning engines to create a repeatable deployment process for machine learning models.

Transcript

1. Holiday Extras Case Study: Employing Google Cloud Machine Learning Engine to develop models in production REBECCA VICKERY | DATA SCIENTIST @vickdata
2. Rebecca Vickery Data Scientist, Holiday Extras @vickdata medium.com/rebeccavickery
3. Scaling machine learning with Google Cloud
4. A data science journey @vickdata
8. COLLECT, MOVE, STORE
9. COLLECT, MOVE, STORE TRANSFORM, AGGREGATE, LABEL
10. TEST, LEARN, PREDICT TRANSFORM, AGGREGATE, LABEL COLLECT, MOVE, STORE
11. Data science team purpose: To optimise customer experience and business performance using automated decision making leading to greater profitability
12. NOW Simple Machine Learning FUTURE AI, Deep Learning
13. Deploying machine learning is hard @vickdata
14. Tools - Data Science Open source Lack software development expertise Mainly python c Flaticon
15. Tools - Software Engineers Different tools Lack ML/Data expertise Mainly javascript c Flaticon
16. Data science process The wrong kind of independence c Flaticon
17. People Small data science team Science + software experts are rare c Flaticon
18. Two types of deployment
19. Bespoke Solutions “Ideas are worth nothing unless executed”, Derek Sivers c Daniel Moyo
20. Unused Models Many models never make it to production “Ideas are worth nothing unless executed”, Derek Sivers
21. Time to model deployment Model development = days to weeks Model deployment = weeks to never! “Ideas are worth nothing unless executed”, Derek Sivers
22. The Google Way @vickdata
23. Machine Learning Engine A technical solution but also a strategic solution c Google
24. The right kind of independence c flaticon Data Scientists can use preferred tools
25. The right kind of independence c flaticon Repeatable process for deployment of most models
26. Faster time to production c flaticon Fully managed service
27. Faster time to production c flaticon Fast response from Google Cloud Support
28. Extensibility c flaticon.com Multiple models and versions
29. Extensibility c flaticon.com Regularly release support for newer versions of tools
30. c Flaticon __init__.py task.py setup.py model.py Model Package
31. Repeatable, reusable process __init__.py task.py setup.py model.py Model Package
34. Collaborative Project
35. ML Proxy (bespoke ML microservice)
36. Monitoring - model performance
37. Monitoring - ML Engine performance
39. Three key benefits @vickdata
40. The right kind of independence
41. Collaboration for data scientists
42. Data scientists deliver more value
43. Thank you @vickdata