How to Handle the Training Model Hangover

We are using ML when we need external data to reach a working product because it would be impossible to solve it with the regular for/if/loops. What are the next steps? Moreover, what about Test, Release, and Deployment?

Our Organizations may be “data-driven”, but now the impact we seize is bigger than we can imagine. No Link Found If you are using an ML component, misused/dirty/problematic data will affect not your internal reports as before… but your application deployment and quality of service.

Let’s discuss some AI implementation stories (its advantages/problems) finding common mistakes and future challenges for such a hyped theme.

Transcript

1. How to Handle the Hangover? Pratik Parmar CLOUD AND ML
2. @hackyroot
3. How to Handle the “Training Model” Hangover? Pratik Parmar CLOUD AND ML
4. @hackyroot I Am Pratik Parmar I am here to bore you with Machine Learning! Hello! You can contact me at @hackyroot
5. @hackyroot
6. @hackyroot The Timeless Conflict @hackyroot IT Ops Dev THE TIMELESS CONFLICT
7. @hackyroot The Timeless Conflict @hackyroot DevOps
8. @hackyroot The Timeless Conflict @hackyroot Product Dev IT Ops InfoSec QA
9. @hackyroot GOOGLE AI MAKES PHONE CALLS??!! CAN YOU DO SOMETHING LIKE THAT USING TENSORFLOW STUFF? IT WOULD REALLY HELP US GET MORE FUNDING! - PERSON A, VP OF PRODUCT, LAST WEEK “
10. @hackyroot WAIT… CAN YOU ALSO DO THAT IN A CONTAINER USING BLOCKCHAIN? - PERSON B, ONE DAY LATER “ * PERSON B WAS APPOINTED VP OF PRODUCT
11. @hackyroot SO OBVIOUS! WE WILL CREATE AN AI DEVOPS CLOUD NATIVE APPLICATION ON KUBERNETES USING BLOCKCHAIN THAT THE USER WILL EXPERIENCE WITH A VR HEADSET? A TRULY DISRUPTIVE DIGITAL TRANSFORMATION STRATEGY! - PERSON C, YESTERDAY “ * PERSON C IS THE NEW CEO
12. @hackyroot MAKE COMPUTERS CAPABLE OF DOING THINGS THAT WHEN DONE BY A HUMAN, WOULD BE THOUGHT TO REQUIRE INTELLIGENCE AI
13. @hackyroot MAKE MACHINES FIND PATTERNS WITHOUT EXPLICITLY PROGRAMMING THEM TO DO SO AI MACHINE LEARNING
14. @hackyroot The Timeless Conflict @hackyroot
15. TRADITIONAL DEVELOPMENT REVOLVES AROUND IF/ELSE, LOOPS & FUNCTIONS RETURNING A VALUE
16. ML DEVELOPMENT REVOLVES AROUND HYPERPARAMETERS & DATA PIPELINES
17. @hackyroot The Timeless Conflict @hackyroot Product Dev IT Ops InfoSec QA Data Scientist ML Engineer
18. @hackyroot The Timeless Conflict @hackyroot YOU DO YOUR BEST TO IMPROVE
19. @hackyroot The Timeless Conflict @hackyroot WHILE YOU ROCK & THROUGHPUT INCREASES, A DATA SCIENTIST BUILDS A MODEL
20. @hackyroot The Timeless Conflict @hackyroot LOCAL DEVELOPMENT
21. @hackyroot The Timeless Conflict @hackyroot CSVs ON A LAPTOP
22. @hackyroot The Timeless Conflict @hackyroot MODEL BUILDING TAKES TIME.
23. @hackyroot The Timeless Conflict @hackyroot SHOW TO THE CEO!
24. @hackyroot The Timeless Conflict @hackyroot CSVs ON A LAPTOP WE ARE AN AI COMPANY!
25. @hackyroot The Timeless Conflict @hackyroot NOW DEPLOY THIS!!!
26. @hackyroot The Timeless Conflict @hackyroot IT Ops Dev ML / DS THE TIMELESS CONFLICT PART II - THE RETURN OF AI
27. @hackyroot The Timeless Conflict @hackyroot PROBLEMS I HAVE SEEN (SOME I MAY HAVE EVEN CAUSED)
28. @hackyroot The Timeless Conflict @hackyroot DOING WHAT YOU’RE NOT GOOD & HIRED FOR...
29. @hackyroot The Timeless Conflict @hackyroot FEATURE ENGINEERING WON’T EASILY SCALE
30. @hackyroot The Timeless Conflict @hackyroot FROM CSVs TO EVENTS
31. @hackyroot The Timeless Conflict @hackyroot MODEL DRIFT & FRAGILITY
32. @hackyroot The Timeless Conflict @hackyroot ONBOARDING
33. @hackyroot The Timeless Conflict @hackyroot HARD TO MEASURE
34. @hackyroot The Timeless Conflict @hackyroot CULTURE CHANGE
35. @hackyroot The Timeless Conflict @hackyroot CONTINUOUS DELIVERY
36. @hackyroot The Timeless Conflict @hackyroot DEMO GOD, PLEASE BE KIND....
37. @hackyroot The Timeless Conflict BEFORE I RUNAWAY, QnA TIME...
38. @hackyroot I’M LOOKING FOR NEW OPPORTUNITIES AS A DATA SCIENTIST! TALENT ALERT!
39. @hackyroot Thank you the great JD Hancock for the amazing pictures! http://photos.jdhancock.com
40. @hackyroot Thank you Thiago de Faria for giving me an inspiration to talk on this topic!
41. @hackyroot You can find me at: @hackyroot hackyroot@outlook.com Any more questions? Thank You!