I (John Mount) am recommending a book that I just started reading.
The publisher Manning recently reached out to me and asked if I would accept a free copy of Effective Data Science Infrastructure by Ville Tuulos in exchange for considering helping to promote it. No obligation to promote it, but I want to disclose I got a free e-copy [though, being a Manning author, they typically will give me a free e-copy of any of their publications].
On to the book: I opened it and ran right into this figure.
Effective Data Science Infrastructure by Ville Tuulos
I actually laughed out loud (in agreement and joy).
The above is literally what Dr. Nina Zumel has been teaching in her role as VP of Data Science Practice at Wallaroo. The parts the data scientist doesn’t care about have to be done well, or there is no point. She represents the data science point of view in an ML-ops company, helping them to design a product that is friendly and useful to both data scientists and ML engineers.
From the Wallaroo website:
Machine Learning is Hard
Deploying models doesn’t have to be.
Of course the above is the engineer’s point of view, which is why they need Nina (to whom machine learning is not in fact hard).
In our book we teach that one needs domain empathy to work with others on their terms. This new book is a chance to learn empathy for operations.
So if you need a tool for model management and deployment please contact Wallaroo, and if you want to learn the topic please check out Effective Data Science Infrastructure by Ville Tuulos.