I guess I'll have to build radar later

I was recently offered a job building radar at an Australian defence contractor, and have been pondering whether to pursue software or hardware over the next few years.

On the one hand, hardware is clearly a much more enduring skillset now that we have LLMs. I dropped out of my computer science degree because it’s so easy to teach myself everything I might need to know about computer science. I finished my robotics engineering degree because there’s a skillset there that’s much harder to pick up from the internet alone, and only really possible to finetune through trial and error and learning from others in real life. If I don’t go into a hardware career now, I’ll have a lot of catching up to do in 10 years. And I suspect that at some point in my career all the interesting tech work will move to hardware, so I’ll probably want to do that catching up.

On the other hand, software is certainly not over yet - AI is clearly going to bring a LOT of value to the economy in the next few years. The level of dopamine in software is higher than ever before now that it’s so much faster and easier to write code. And the career rewards feel much closer than hardware, which will likely take a while to pan out in the same way AI is now. I would probably be crazy to miss the opportunity.