Accepting My Fate as a Millennial Software Engineer

In my humble opinion, younger engineers are not well served by the notion that compilers, operating systems, network stacks and low level code are “solved” problems.

I’d argue this from two angles. First, are people still complaining about problems with operating systems and languages? Second, is innovation still occurring? I’d argue that the answer to both questions is a resounding yes.

Even in something as simple as network stacks, the relatively recent explosion of mobile devices and the IOT has driven change. How about low level languages? Rust has just shown up on the scene showing that low level programming wasn’t “solved” by C/C++. And I think we can all still think of big problems with our favorite operating system.

Last, consider a far older technology - the automobile. Are exciting things still happening with cars? Sure, innovation has greatly slowed down, but the electric car has the promise of reinventing the entire industry - and that’s innovating the most fundamental mechanical components like the drive train and energy source.

I think computing technology, being way younger than the automobile, still has room for big improvements at all levels of the stack.

Commercial software will always be a "black box". Many, many companies use commercial software. This is not really anything new. I will admit that the complexity has grown so much that it's nearly impossible to understand any significant software system in its entirety. That's not new, either, just more prevalent than it used to be.

Quantum Computing is so new, it's in the realm of op-code level abstraction, maybe even processor architecture.

goto grad school and write papers. you can explore topics that interest you more deeply, and while you may not know everything about a large topic like ML, you can focus in on a specific sub-topic and become the world's leading expert.

https://www.datarobot.com/blog/a-primer-on-deep-learning/

You probably know most of the information in this post about ML, but this picture at the top shows what researchers are trying to discover about ML.

Basically, researchers are inspecting the middle layers of neural networks to determine patterns. It's kinda a backwards approach to understanding nn's, as we already can use them without understanding them, but this should show that there's opportunity to really dive into these "deep" questions in academia.

I think this is the actual paper, but it seems to have a wider topic:

http://www.cs.toronto.edu/~rgrosse/icml09-cdbn.pdf

I strongly encourage you to go off the beaten path with the three ideas you have there - but spend time looking for places they have been done. It does feel at times all we’ve done is build all these things up for webapps.