I Got Fired Because of AI, But I Still Consider Myself an Engineer of the Future

Over the past few months I have completely rebuilt my approach to development. I used to write code myself. Now I strategically manage neural networks.

I used to spend time on architecture, reading documentation, and understanding other people's code. Now I spend time choosing the right model and debating on Twitter which one is the real game changer.

I bought all the most expensive AI development tool subscriptions — five hundred dollars worth. What suckers those people are who still haven't done this. I follow all the AI tool vendors and read their posts. If someone writes that their tool replaces a mid-level engineer, I see it first. You have to think one step ahead of the market.

I used to worry that I didn't understand some complicated part of a system. Now I haven't understood what the neural networks wrote for me in a long time. But that's fine — I read on Twitter that reading and understanding code is passé. The modern engineer doesn't dig in, he orchestrates with agents. If the code compiles, everything is fine. If it doesn't compile, you need to rephrase the prompt.

I no longer create tasks in Jira — I do it through Claude. Yes, it takes longer, but at least I'm on trend. First I ask it to formulate the task, then I clarify the context, then I ask it to rewrite more structurally. An hour later I have the perfect ticket.

Colleagues started asking strange questions: «Did you write this yourself?» or «Do you understand how this works?». I consider this old-fashioned thinking. What matters is not understanding — what matters is delivering. And if something breaks, you can always say the model hallucinated a bit. We all understand that's temporary.

Sometimes I open a pull request and can't explain why it's written the way it is. But at least it all looks modern: abstractions, layers, generic types, a couple of new libraries. Looks solid. If anything comes up — you can always ask the model to explain what it meant.

The hardest part is calls. Before, I could immediately explain why I chose a particular solution. Now I ask the neural network in advance to generate an explanation for the architectural decisions it came up with. It comes out convincingly. Sometimes I even start believing I came up with all of it myself.

Everyone says that anyone who doesn't use AI will be fired. I was fired precisely because of AI — because I started writing code of terrible quality. But I'm working for the long game. Yes, production went down. Yes, automated tests suddenly started testing the wrong service. Yes, functions appeared in the codebase that nobody called. But that's the price of transformation.

After being fired I started building my own startup. Naturally, I didn't write the code — I assembled it with Claude Code in three days. I want to build a business and earn $10,000 a month. But there's a small problem right now. I have one critical bug that prevents the application from launching. The neural networks couldn't fix it, and I don't know what to do in a case like this. I'm currently looking for information on Twitter about how to proceed. So the startup is on hold for now.

People tell me I've become dependent. I prefer the formulation «augmented myself with tools». But sometimes, deep in the night, when yet another assistant freezes on the words «Thinking…», I stare at the screen and think: maybe I should actually figure it out myself? But then I remember that reading and understanding code is passé.

And if tomorrow another tool comes out for a hundred dollars that promises to write everything perfectly — I'll buy it, of course. You can't fall behind.

Because what if this time it really does work perfectly.

Disclaimer: This is just an ironic look at what's happening. Don't take it seriously — or do. Your call.