Hey {{first name | there}}. I posted this on LinkedIn and so many engineering leaders shared their opinions.
In today's career notes:
Why 2 years of "AI will automate infrastructure" just quietly failed
The skill gap AI assistants are creating in engineers who lean on them
What companies will actually be desperate for by 2028
🧗THE EDGE: The DevOps job market is about to collapse
You might hate me for saying this.
But the DevOps engineering job market is about to collapse.
And then it's going to explode.
There's just a big catch.
The 2026-2028 window is going to separate the engineers who survived on AI prompts from the ones who actually mastered the fundamentals.
Here's what I'm seeing across 383+ coaching calls with engineers on 6 continents.
The last 2 years, in one sentence
Companies bet that LLMs could automate their entire infrastructure.
They can't.
AI is genuinely great at generating a boilerplate Terraform script or a basic Dockerfile.
What it doesn't understand is the nuance of a production environment at scale.
It doesn't know why a specific latency spike is killing your database connections.
It doesn't know how to navigate a complex incident when the networking gets messy and three systems are lying to you at once.
Meanwhile, a quieter problem has been building
Engineers who leaned too hard on AI assistants are losing the ability to architect systems from first principles.
↳ They can't debug a broken CI/CD pipeline without a prompt.
↳ They don't understand the "why" behind their Kubernetes configs.
↳ They freeze when the AI-generated solution doesn't work in the real world.
I've watched this show up on calls. Someone can walk me through what the AI suggested. They can't tell me why it was the right call, or what they'd do if it wasn't.
That gap doesn't show up on a resume. It shows up in a postmortem.
Where this is heading by 2028
Companies aren't going to be hiring "YAML engineers."
They're going to be desperate for engineers who can:
↳ Architect multi-region infrastructure that handles 10M+ requests/second.
↳ Build internal developer platforms that actually improve velocity instead of adding a layer of ceremony.
↳ Optimize cloud spend without quietly wrecking performance.
↳ Go kernel-level when something breaks and the dashboards go quiet.
To be clear, I'm not anti-AI
I use it every day. So does everyone worth learning from.
This isn't about rejecting the tools. It's about what happens to the engineers who let the tool replace the thinking instead of accelerate it.
The ones who spend these two years actually understanding networking, distributed systems, and resilient infrastructure design are going to be diamonds.
Everyone else is going to be competing for the same shrinking pool of entry-level roles, and the bar for those roles is only going up.
🧠PRESENTED BY EVERYTHINGDEVOPS
Divine breaks down platform engineering, Kubernetes, AI, and the career lessons, minus the big-tech gatekeeping in a format you can actually scroll on your break.
Follow us @divinelovescloud
🔭VISIBILITY STACK
I am on Instagram now – I PROMISE TO MAKE FUN & TECHNICAL VIDEOS 🥳
Agentic AI Bootcamp & Certification: Build & Deploy Multi-Agent Systems – I plan to enroll in this bootcamp by Priyanka Vergadia (Ex- Senior director @Microsoft, @Google) in January. Check it out.
AI Was Supposed to Kill Engineering Jobs. It Didn't. – Here Prince share why AI can’t take your Job just yet!
Stay sharp. Stay visible.
Divine Odazie
CEO, EverythingDevOps
✅ACTION ITEMS FOR YOU
Your homework. One question. Answer it honestly.
Which side of that gap are you building toward:
The engineer who understands the "why," or
The one still one prompt away from being stuck?
Reply and tell me. I read every one.
👀
Your move.