Gentlemen IT experts! I understand your resentment towards me and your attempts to declare me "my mother's vibe coder".
Yes, today almost anyone can build an MVP, website or dashboard in an hour. And what? Do you seriously think this is where AI's capabilities end?
I have one simple question for you: why did you decide that a person who can develop an application using AI cannot use the same AI to build an infrastructure for him and automate its operation? I simplify my posts on purpose. I'm not going to write technical documentation every time for those who consider themselves professionals.
But since we're talking about "what will you do when you hit 50k users", let's talk a little about my infrastructure.
So what do I have for deploying applications in production: Docker Compose: a container layout that can be restored. Traffic/Caddy: HTTPS, hook. PostgreSQL: continuous data storage. N8n: Your own performance in your domain, automation of processes and integrations. Redis+ queue: keeping the cache, background work. GitHub: CI/CD, automated tests and deployment. A: Tracking app errors. Prometheus + gharana: indices and monitoring. Backup: Restore data after disasters.
Our n8n workflow is set up, webhooks and credentials are connected. For critical operations, impotence is provided, i.e. protection against repeated execution of the same business operation. And now about the "s*** of AI codes, then who will support it? " You seem to be missing one funny detail. AI can not only generate code, but also test it.
There are special skills and tools of agents: Code review: looking for errors, anti-patterns, code duplication. Security Review: Vulnerability Testing, OWASP, Confidential Leaks. Architecture review: analysis of architecture, dependencies, scalability. Generating tests: Generating unit, integration and E2E tests.