I find the problem before choosing the solution.
Senior platform and infrastructure engineer with three decades across production systems. Databases, Linux, AWS, automation, security, resilience and regulated technology delivery — now combining that systems experience with AI-assisted engineering, Kubernetes and GitOps.
Systems thinking over product loyalty.
The technology changes. The engineering method doesn't: understand the system, distinguish evidence from assumption, expose hidden dependencies and failure modes, then choose the most supportable path.
Established experience. New tools on top.
Kubernetes and GitOps are additions to a broad production infrastructure background — not a replacement for it.
Production engineering
Cloud and infrastructure work spanning investigation, design, implementation, controlled production change and operational handover.
Modern platform layer
Hands-on labs focused on cloud-native orchestration, GitOps and governed AI-assisted operations.
AI is a force multiplier.
AI is now a major part of how I engineer. Not just autocomplete, and not autonomous production decision-making: I use it to retrieve, connect, analyse and challenge evidence at a speed that materially changes the engineering feedback loop.
Evidence + domain expertise + AI reasoning
Enterprise information retrieval, structured prompting, Python-assisted analysis and infrastructure domain knowledge combine into something more useful than a chatbot: a reasoning and evidence accelerator. I use AI to reconstruct fragmented context, explore unfamiliar systems, test hypotheses, generate and review implementation options, diagnose failures and accelerate documentation.
Control principle: AI accelerates reasoning and implementation. Source tracing, human verification, change control and engineering governance stay in the loop.
✓ Argo CD reconciliation / self-healing tested
→ Connect Git → Argo → K3s
→ AI diagnosis → proposed PR → approval → Argo remediation
→ Floci for fast Terraform / AWS testing
The builds are evidence, not the identity.
The original peterpain.com demos live here now: practical environments for testing platform engineering, local AI and automated recovery ideas.
Self-healing CI/CD
Failure → diagnosis → minimal change → controlled redeploy → verification. Exploring how AI can shorten recovery without bypassing Git or human control.
Working stackLocal AI Stack
Local inference and agent-assisted engineering workflows for rapid experimentation with low cloud dependency.
Working platformHomelab Platform
Proxmox, networking, storage and platform services: real infrastructure used as a safe engineering laboratory.
K3s + Argo GitOps Lab
Kubernetes, Helm and Argo CD, building toward an auditable AI-assisted remediation loop.
Floci AWS Test Lab
Fast local AWS/Terraform integration testing as a safe proving ground for infrastructure and AI-generated changes.
The self-healing idea, evolved
The original site led with “pipelines that fix themselves.” The idea stays — but with the control model made explicit.
Platform engineering, SRE and AI are converging.
I'm not actively job hunting today. I'm building toward the next chapter: senior platform, SRE and AI-infrastructure work where deep production experience and strong AI-assisted engineering reinforce each other. UK or remote, when the fit is right.