MARCUS PATMAN
Expertise · 05

Cloud
Platforms

AWS, GCP and Azure in production since 2015 — with the scars to prove it: migrations that couldn't blip, bills that had to make sense, and Kubernetes that had to stay up.

What I mean by it

Direct answer Cloud platform work is making the big three behave like infrastructure you can trust: everything in Terraform, every workload migratable, every cluster observable, and every bill explainable to a CFO.

Multi-cloud isn't a buzzword on this page — at ConstructConnect it was the daily reality of moving workloads between AWS, GCP and Azure without downtime budgets to spend.

Production record

3Clouds in production
200+Workloads → GCP
50%Faster provisioning
99.9%K8s uptime

Employer-stated role outcomes, 2015–2025 · Terraform · Kubernetes · GitOps (ArgoCD)

Security in the cloud

Cloud credentials are the new root password, so scanning for them became part of my toolchain: Secret-Scan audits repositories for AWS keys, GitHub tokens and API secrets at ~51,000 files per second, with obfuscation detection — because leaked credentials don't wait for a change window.

Common questions

Is multi-cloud ever actually a good idea?

Not as a default — the tax is real. It earns its keep when a business constraint demands it (acquisitions, data residency, an exit clause). I've run it where it was required and kept the blast radius small: consistent Terraform modules, one delivery pipeline, per-cloud abstraction only where the platforms truly differ.

How do you keep cloud bills sane?

Tagging enforced by policy, budgets with alerts before the invoice, and regular rightsizing passes — the 45% deployment-speed work in my architect era came with cost accountability attached. Boring hygiene beats exotic FinOps tooling.