Thank you WorkOS for sponsoring! Check them out at: https://bigboxswe.dev/workos

Sources
[0] Self hosted projects list: https://github.com/awesome-selfhosted/awesome-selfhosted 
[1] All of Google's SRE books: https://sre.google/books/ 
[2] The SRE Book I read specifically: https://sre.google/sre-book/table-of-contents/ 
[3] Devops course: https://www.boot.dev/paths/devops?tech=python-golang 

Music:
@massobeats : https://www.youtube.com/watch?v=egF4ek4MHOw

BigBoxSwe's resource offers insights, tutorials, and resources for software developers and technology enthusiasts. Readers can learn about coding best practices, software architecture patterns, and emerging technologies. With articles, tutorials, and code samples, BigBoxSwe provides  guidance and expertise for building software applications and honing technical skills.

bigboxSWE

A video transcript arguing that AI-assisted 'vibe coding' is fundamentally unsuitable for infrastructure and reliability engineering. The core argument is that UI features have tight feedback loops (caught in dev or tests), but infrastructure failures manifest as multi-hour production outages with no clear root cause. Recent real-world outages (Grok uploading git repos, Railway losing Google Cloud API access, COD outage without postmortem) are cited as evidence that infrastructure complexity cannot be automated away. The piece recommends learning SRE fundamentals via Google's SRE handbooks, Boot.dev's DevOps course, and self-hosting projects. Sponsored by WorkOS for enterprise authentication.

Why you can't vibe code real software

Coder is a junior who knows the syntax of one language and the API of a couple of libs/services. But to advance their career, they also need to begin immersing themselves in fundamental knowledge. AI is currently aiming to fill this niche. And yes, it comes preloaded with syntax documentation for virtually all known languages and libraries.
Software engineer identifies a problem and assembles the architecture of the entire stack to address it. They implement the necessary algorithms and patterns for the codebase, databases, security systems, control, monitoring, and deployment. And at the lowest level -writing code, scripts, db queries, tests, docs.
AI can be a powerful tool at each stage, but it can’t do all this without orchestration and verification of results.

But I have already and am selling it.

but that is why there are companies like replit and rantir emerging

The infrastructure angle is the sharper version of this. Vibe coding shines where the feedback loop is tight — the code compiles and passes the test, or it doesn’t. Infra failure is the opposite: it shows up hours later as a multi-service outage with no clean stack trace, and the fix depends on context that was never in the prompt. A model can scaffold a Terraform file; it can’t hold the model of how your VPC, IAM, and retry budgets interact under load. That’s the 20% you can’t automate away — not because it’s hard to type, but because it’s hard to reason about, and the failure signal arrives too late to vibe your way through it.