How are you doing machine learning, especially in the “cloud”?

Assuming you do it professionally or are involved in a team or organization that use machine learning, how do you go about your machine learning projects? What is your workflow? What tools do you use and why? What tools don't you use and why not? How did you evolve and come up withe way you're doing things? How is your team organized if you have a team? How do you collaborate and hand-off each piece of the puzzle? How do you handle data, develop your models, track experiments, deploy and manage your models, etc.

Which cloud provider do you use? Do you use their "AI" offering, or do you use them purely for the infrastructure (compute, storage, containers, etc)?

What's onboarding a new team member like? What do they have to learn? Best practices you came up with or follow?

I like interacting in threads such as this one[0] or ones addressing frictions.

- [0]: https://news.ycombinator.com/item?id=25938654