Most full stack developers have breadth but not depth. It's incredibly time-consuming to have a complete understanding of the entire stack and no matter what you do, even if you live and drink code and do it all the time you will not be a complete engineer, there is just too much and you will never retain it all. Rather than try to be this ninja superstar that start-ups all look for these days its more realistic to be able to work full-stack but specialize in a particular field. Machine learning and a solid academic background in stats, math and understanding of concurrency will get you just as far as a guy who knows React, Sass, HTML5, Python, Ruby, Go, MongoDB, MySQL, AWS, Redis, Celery, Kafka and Hadoop.
I recently move from a full-stack position into a more focused position. I felt like I was "jack of all, master of none" which really irked me. The time I was given for training was good, but it had to be split among all the different layers of the stack which meant I never really got an opportunity to get deep into any particular layer.
> Most full stack developers have breadth but not depth.
I have to disagree with this, its a generalisation at best. In my experience the "full stack" developers have had a much deeper knowledge of a subject then the "specialist". This is simply because the layers in the stack are not as discrete people would like you to think and the full stack developer has a much better understanding of why and how something is.
Context matters. Small startups will usually benefit a lot more from an engineer who can be flexible and create a lot of immediate value as opposed to an ML expert, at the outset. Large corporations are a different story of course.
