Senior SRE & Platform Engineer — Kubernetes Observability, AI-assisted Tooling & Cloud Operations
_{Abingdon-on-Thames, United Kingdom}

5 years in site reliability and platform engineering, working on the platform that runs seismic interpretation and subsurface planning tools for energy companies in production across Europe, the Americas, and Asia-Pacific.

I build tooling that makes Kubernetes observable and incidents shorter. My current focus is AI-assisted SRE tooling — deterministic CLI tools paired with Claude Code reasoning skills.

Project	Description	Status
kubectl-sentinel	10-section Kubernetes health checker. Nodes, pods, workloads, probes, services, HPAs, PVCs. Structured JSON + HTML output. Works in CI.
incident-triage	PagerDuty alert → root cause → causation chain → fix plan. Deterministic correlation engine. Python stdlib only.

Kubernetes Observability    ██████████████████░░   Prometheus, Thanos, Grafana, Kiali, Istio
Incident Response           █████████████████░░░   PagerDuty, causation chains, root cause classification
Platform Automation         ████████████████░░░░   Helm, ArgoCD, CI/CD pipelines, tenant provisioning
Cloud Operations            ██████████████░░░░░░   GCP, Azure, multi-region, cost optimisation
Security & Compliance       ████████████░░░░░░░░   RBAC, TLS, pod security, secrets hygiene
AI-assisted Tooling         ████████████░░░░░░░░   Dual-layer CLI + LLM reasoning pattern (building)

• The dual-layer pattern — a deterministic CLI layer that works at 3am in CI, paired with an AI reasoning layer that explains why. Separating them gives you both portability and intelligence.
• Severity as an exit code — 0/1/2 makes health checkers CI-composable. Think of them as HTTP status codes for shell tools.
• Classify before you recommend — OOMKill does not mean "raise the limit". The root cause determines the fix.
• Building for the 3am reader — every design decision in an ops tool should be made as if the reader has been awake for 3 hours and needs to act in 5 minutes.

→ GreenRanger-IT/learning-journal — architecture decisions and engineering thinking from building production SRE tooling

Working toward CKA → CKS → CNCF contributor. Open to collaborating on Kubernetes observability and platform engineering tooling.