Cloud architect based in Dublin. Writing about Azure, Kubernetes, CI/CD, and structured AI-assisted development.
8-article series on building an Internal Developer Platform with Backstage where AI is a first-class citizen — from service catalog to incident response.
8-article series on the architecture decisions that determine whether your AI project survives contact with real users.
7-article series on securing modern cloud systems with post-quantum cryptography — QuantumVault, QuantumAPI EaaS, and QuantumID.
10-article series teaching structured AI-assisted development using the ATLAS and GOTCHA frameworks.
Five articles, three frameworks, one team. Time to put it all in one place — the templates, the metrics, and the decision rules that actually stuck for the QuantumAPI team after 6 months. Copy, adjust, ship.
Four articles saying 'use this framework, it works'. Time for the uncomfortable part: none of them is a silver bullet. Each has a sweet spot and a cost — and teams that stack all three without thinking end up slower than they started.
SDD gave us small batches. GOTCHA gave us useful feedback. But when production broke at 2am, the on-call engineer could not explain the AI-generated code they were looking at. Time to use AI as a tutor, not just a generator — and finally close the Third Way.
SDD gave us small PRs. But the AI reviewer still left 22 comments on a clean 5-file diff, and 19 were noise. The team learned to ignore it. Here is how GOTCHA turns a crying-wolf reviewer into one that gets read.
The 47-file PR from the last article was not the AI's fault. It was ours — we gave it a prompt, not a spec. In this article, the same team rescues the same feature using Spec-Driven Development, and watches lead time drop from 6 days to 1.