AI is the buzzword while ML is the underlying component... but when do we use ML? To solve problems that machines can find patterns without explicitly programming them to do so. But do you have a team building an ML model? How far are they from the IT team? Do they know how to deploy and serve that? Testing? And sharing what they have done? That's where a devops mindset comes in: reduce the batch size, continuous-everything and a culture of failure/experimentation are vital for your data team! In the end, I will show how the workflow of a data scientist can be on the real life with a live demo!
12. @thiagoavadore#codemotionBerlin
GOOGLE AI MAKES PHONE CALLS??!!
CAN YOU DO SOMETHING LIKE THAT USING
THAT TENSORFLOW STUFF?
OUR USERS WOULD LOVE THAT!
PERSON A, VP OF PRODUCT
The real goal? A great PR
HYPE WARS - REVENGE OF AI
16. @thiagoavadore#codemotionBerlin
WHY DON’T WE CREATE AN AI DEVOPS
CLOUD-NATIVE APP ON K8S USING
BLOCKCHAIN BACKED BY IOT DEVICES THAT
THE USER WILL EXPERIENCE WITH A VR
HEADSET? A TRULY DISRUPTIVE DIGITAL
TRANSFORMATION!
HYPE WARS - THE ULTIMATE MEDIUM READER
17. @thiagoavadore#codemotionBerlin
WHY DON’T WE CREATE AN AI DEVOPS
CLOUD-NATIVE APP ON K8S USING
BLOCKCHAIN BACKED BY IOT DEVICES THAT
THE USER WILL EXPERIENCE WITH A VR
HEADSET? A TRULY DISRUPTIVE DIGITAL
TRANSFORMATION!
PERSON C, appointed CEO and master-of-all things
Gets funded and moves to the Bay Area
HYPE WARS - THE ULTIMATE MEDIUM READER
42. @thiagoavadore#codemotionBerlin
THE DATA PIPELINE & ML EXTRA PROBLEMS
▸ DS != software engineers
▸ Data pipeline won’t scale
▸ Local development, local data
& lack of versioning
▸ Lot of rework
▸ Packaging, deploying &
serving??!
43. @thiagoavadore#codemotionBerlin
THE DATA PIPELINE & ML EXTRA PROBLEMS
▸ DS != software engineers
▸ Data pipeline won’t scale
▸ Local development, local data
& lack of versioning
▸ Lot of rework
▸ Packaging, deploying &
serving??!
▸ Even harder to measure the
efficacy