The Non-Traditional Path to DevOps Engineering
I went from Twitch streaming to running cloud infrastructure. Via hospitality, a failed recruiter side quest, and a traineeship I almost missed. Here's the real story.
Writing
The non-traditional path, certifications, and honest career advice.
I went from Twitch streaming to running cloud infrastructure. Via hospitality, a failed recruiter side quest, and a traineeship I almost missed. Here's the real story.
When you walk into a new organization you see everything that could be better. That is the easy part. Making it actually better, without breaking the team that has to live with it, is a separate skill entirely.
19% of companies admit their AI screening tools reject qualified candidates. If you have a non-linear career path, you're probably one of them.
Everyone measures Applied Skills against certifications and finds them lacking. That framing misses the point. The real value is not the badge. It is free structured hands-on time with Azure services your job will never let you touch.
Certifications are not proof of competence. They are not efficient learning tools. They are a ticket past a broken filter, and the only rational way to use them is to stop pretending they are anything else.
At a big company you're the Terraform guy. At a small company you're the Terraform guy, the network guy, the security guy, and the printer guy. That's a feature, not a bug.
What actually works, what doesn't, and what nobody talks about.
After a year of using AI tools daily in production IT work, here's what genuinely helps, what's overpromised, and what nobody talks about. No hype, no fear, just what I've seen.
Microsoft shipped Agent Framework 1.0 on April 3, 2026. I built a multi-agent Azure FinOps dashboard to stress-test it against real Azure APIs. Here is what I found.
88% of AI pilots never reach production. 40% of agentic AI projects will be cancelled by 2027. Here's what the data says about why, what actually works, and what might change.
Claude Code is not replacing engineers. It is exposing a split in engineering labor that the industry has been quietly avoiding, and the side of the split you have been working on decides what the next five years of your career look like.
We used to debug by reading logs and forming hypotheses. Now we paste errors into AI and hope for the best. The fix usually works. The understanding is gone, and that is a bigger problem than it looks.
AI-generated code is dangerous not because it looks bad, but because it looks good. That inversion breaks the instincts code reviewers have relied on for twenty years, and almost nobody is talking about what that means for how we ship software.
Landing zones, Terraform, compliance, and lessons from production.
Landing zones, Terraform modules, compliance baselines, cost optimization - what actually matters when you're building Azure infrastructure for real organizations, not certification exams.
Every workload deployment followed the same pattern but took hours to assemble. So I built a wizard that generates production-ready Terraform from clicks, with CIS compliance and AI review baked in.
Embedding CIS Azure Foundations Benchmark controls directly into Terraform modules using preconditions and check blocks. A module library on top of Azure Verified Modules.