Become a Cloud Engineer in 2026

Added: 2026-05-18

[00:00:00]I've been out of college for almost 8 years. So, to some of you, that might make me unk. But, given how much AI can do today, I feel anxious if I were to graduate today. But, the good news, we still hire early career engineers and builders. If you're a new grad or someone hoping to transition in their careers, here's some practical things you should know working on the cloud in the age of AI. Coding is becoming increasingly easy for AI to accomplish end-to-end. So, it's more important than ever to know what to build and why it's built that way. Here's a main principle for juniors brought to you by Mark Brooker, distinguished engineer at AWS who played an integral role building Lambda, Aurora DB SQL, and other crucial AWS services. The junior path engages much earlier with economics, product, and people, has less emphasis on the practice of the craft of programming, but more emphasis on the deep technology and science behind the systems we are building. Although, a lot of the work that consumed a junior engineer's typical first year, such as writing boilerplate infrastructure files, wiring up permissions by hand, or debugging mismatched network rules at 11:00 p.m., can now be handed off to an AI agent, your background still requires understanding fundamentals in order to build resilient systems that scale.

[00:01:16]That's where skills shift. The best juniors make their senior engineers faster. They ask research questions, they test assumptions before escalating, they can take an ambiguous task, use a coding agent to generate a first draft, and then review it against the fundamentals: security, reliability, cost, performance, and maintainability.

[00:01:36]I remember when Opus 4.5 released, our head of product for EKS said, "Wow, this model is essentially a K8s expert." That doesn't mean the model replaced all platform engineers, it just empowered them to focus more on the hard problems specific to their system. So, the question is, if AI can already generate a Kubernetes manifest, Terraform module, Lambda function, CI/CD pipelines, what are you supposed to be good at? Let's use a real-world example. Say you're asked to add login to a service. AI can generate that in 5 seconds, but this is not production engineering. The questions you should be asking and knowing the answers to are what information does someone need when this fails? Can we trace this request across services? Are we logging sensitive customer data? Is this log easily searchable? Are we logging enough to debug the issue, but not so much that we explode costs? Will these logs drive too much latency in our application? Here are some key resources for you to go off of. Build off the tools and libraries on GitHub and the AWS or AWS Labs GitHub org. Those are built by us and used by real customers. So get a sense what customers care about. You can also use the AWS free plan where you won't be charged unless you upgrade to the paid plan. Your free plan expires after 6 months or when you use up your free tier credits, whichever comes first. You can also test your understanding by taking certifications. For beginners, I recommend passing the AWS Certified Cloud Practitioner exam. It's a 100-level exam designed for people new to the cloud and it gives you the vocab you need to work effectively alongside AI tooling. The next step to showing true understanding is building directly.