[softwaredoug]

I think the ML bubble will become increasingly more realistic, and those that can turn ML into scalable, performant, and maintainable solutions will “win”. Additionally those that focus more on the training data being learned from and it’s quality over the model and how the full product fits together will win...

I think another frontier is knowing how to manage and build organizations that understand data and ML. In the same way software caused traditional engineering orgs to rethink processes, eventually data expertise will infiltrate management. The processes we use to deliver working, well performing data-based solutions will change and cause traditional software orgs to rethink how they deliver work. These processes will be more about running frequent, repeatable experiments and less about iterative feature dev.

So, to sum up:

- ML engineering

- Data scientists that can deliver working code

- managers with data expertise

And I’d emphasize less:

- research, “fancier model” oriented ML and data science

Other random bets:

- office space for remote workers that just want a quiet, private place to work. Boring class A office space converted into single offices rented individually.

- home tech support for remote workers to improve reliability, etc of tech and network experience.

[strikelaserclaw]

which is why i think it's important for software engineers to have some idea about the current state of machine learning.

[cztomsik]

+1 for affordable, come-and-go but private office space