It's great that LLMs provide opportunity for non-software engineers to make tech products, but I wonder how those "vibe-coded" products will fare when faced with actually maintaining the code (and also accounting for tech debt..)
It's not like we're doing very well with maintenance and tech debt in any case. AI might be able to help with that in the future.
Look at your tech stack, go down until you come to the level where things "just work". This is where the maintained software begins. The stuff you fill your docker base images with.
Are you really assuming most of your docker base images won't be AI generated spit in a few months from now? All maintainers are catching on.
But OP is a software engineer, I doubt a non software engineer can turn around a vibe coded app so quickly.
Feel free to judge for yourself what I am. Started in civil engineering, pivoted hard 6 months ago. That gave me a leg up in UI/UX, design experience, all that.
https://www.aizk.sh/Isaac's%20Resume.pdf
Also vibe coding is useless without marketing skills, deployment skills, distribution, social media skills, etc.
> faced with actually maintaining the code (and also accounting for tech debt..)
I guess they'll learn it as they come across it? "Oh Claude, my code is almost like a plate of spaghetti, how can I make it easier to add new features without breaking something else?" "Dear user, here is what technical debt and unit tests mean: ..."
Besides, all of us self-learned programmers mostly learned about those things the hard way as well, by experiencing the real drawbacks of not caring about such things until too late and stuff is already up and running with real users.
Here's the thing: the formally trained programmers learn it that way too. They're just less inclined to admit it.
OP is a software engineer, so this is a software engineer using LLM's to make tech products.
Tech debt of your hobby pet project? It's not like millions of people will be using it.