kubernetes · k3s · gitops · observability
Kubernetes Consultant UK
I help teams get Kubernetes into production properly — with GitOps, observability, and operational confidence. From architecture to runbooks, not just "it deploys".
the problem
Common Kubernetes problems
Kubernetes is running but nobody trusts it
The cluster exists but production is still on VMs or Docker Compose because Kubernetes 'isn't ready'.
Deployments are manual or inconsistent
No GitOps, no automation — just kubectl apply from someone's laptop when a release is ready.
No visibility into what's running
No alerting, no dashboards, no idea if workloads are healthy until something breaks in production.
Resource requests are guesswork
Pods OOMKill in production or massively under-provision, causing noisy-neighbour problems.
Cluster upgrades are terrifying
Nobody knows what will break, so upgrades are skipped until the version falls out of support.
RBAC and multi-tenancy are unclear
Every team has cluster-admin, or nobody has access. Neither is the right answer.
the output
What I deliver
- ✓Cluster architecture review — control plane HA, networking, storage, and ingress design
- ✓Flux v2 GitOps setup — declarative, auditable deployments with automated reconciliation
- ✓kube-prometheus-stack — Prometheus, Alertmanager, Grafana dashboards, and structured alert rules
- ✓Workload review — resource requests/limits, pod disruption budgets, HPA configuration
- ✓Namespace and RBAC design — clean separation between teams and environments
- ✓Ingress configuration — Traefik or NGINX with TLS, routing rules, and rate limiting
- ✓Storage — Longhorn distributed block storage or cloud-provider PV configuration
- ✓Cluster upgrade planning — step-by-step path with rollback strategy
tooling
Tools I work with
I run a production-style 6-node k3s cluster with Flux, Longhorn, Prometheus, Loki, and GitLab CI in my own homelab — not just on a CV. See the full architecture.
engagement models
Typical engagements
Cluster Assessment
1–2 daysArchitecture review, resource sizing, security posture, and upgrade path — written report with findings and a prioritised action list.
GitOps Setup
1–2 weeksFlux v2 installed and configured with your Git repository, Helm releases, image policies, and environment promotion.
Observability Baseline
1 weekkube-prometheus-stack with structured dashboards, alert rules for real signals (not noise), and Loki for log aggregation.
Production Readiness Review
2–3 daysFull workload review covering resource configuration, RBAC, network policies, ingress, and upgrade readiness.
credibility
Why work with me
6-node k3s
3 control plane + 3 workers, HA, kube-vip, MetalLB L2 ARP
116 panels
Custom Grafana dashboards across the cluster
55 alert rules
Structured Prometheus alerts for real signals
Flux GitOps
All cluster state in Git, automated reconciliation
I run a production-style k3s platform in my own infrastructure
6 nodes on Proxmox, Flux v2 GitOps, Longhorn distributed storage, Traefik ingress with TLS, CrowdSec threat intelligence, Authentik SSO, and a full observability stack with Prometheus, Loki, Alertmanager, and Grafana. This runs 24/7, not in a demo environment.
See the full homelab architecture →explore more
Related services
Freelance DevOps Consultant
Broader DevOps support across infrastructure, pipelines, and cloud.
Learn more →GitLab CI/CD Consultant
CI/CD pipelines that build images and trigger Kubernetes deployments.
Learn more →Terraform Consultant
Terraform for provisioning the infrastructure that Kubernetes runs on.
Learn more →AWS Cloud Consultant
EKS and AWS infrastructure for Kubernetes workloads.
Learn more →questions
Frequently asked questions
k3s vs full Kubernetes — which should we use?
k3s is production-ready and I run it in my own infrastructure. For most small-to-mid teams, k3s reduces operational overhead without meaningful trade-offs. Full Kubernetes (kubeadm, RKE2) makes more sense at scale or with specific compliance requirements. I can assess your situation honestly.
Can you help with EKS, GKE, or AKS?
Yes. Managed Kubernetes removes the control plane burden and is often the right call. My experience spans self-managed k3s and managed clusters. The GitOps, observability, and workload patterns are the same across all of them.
We're on Docker Compose — is Kubernetes worth it for us?
It depends on your scale and team. For a 2-person team with 3 services, Kubernetes is overkill. For a team shipping multiple services that need independent scaling, zero-downtime deploys, and proper RBAC, the complexity becomes worthwhile. I won't oversell Kubernetes if it's not the right fit.
Do you do Kubernetes training?
Not structured classroom training — but knowledge transfer is part of every engagement. I'll explain decisions, write runbooks, and make sure the team can operate what's been built without me.
What's Flux v2 and why is it better than just running kubectl apply?
Flux continuously reconciles your cluster state against a Git repository. Every change is tracked, reviewed, and auditable. Drift between Git and the cluster is detected and corrected automatically. It's the difference between 'we think that's deployed' and 'we know exactly what's running and why'.
get in touch
Ready to get Kubernetes production-ready?
Whether you're starting from scratch or untangling an existing cluster, let's talk through what you need.