whoami
I’m Mohamed Belkamel, a Lead DevOps / Platform Engineering Consultant. I work across Kubernetes, AWS, and on-premises environments to improve reliability, delivery, observability, and cost control — following systems all the way from pipeline intent to production reality.
services --list
Find the spend you can safely cut
A baseline of your AWS/EKS usage and cost, then right-sizing, Karpenter/spot strategy, non-prod scale-down, and idle-resource cleanup — with reliability held as an explicit constraint, not traded away. You leave with a prioritised, owned savings plan — and I can ship the Terraform/Helm PRs that realise it, not just hand you a roadmap. Carries a no-findings, no-fee floor.
Make rollouts boring again
Architecture and failure-path review: workload requests/limits/probes, node lifecycle and autoscaling, deployment verification and rollback, observability and incident-readiness gaps — delivered as a risk register you can act on, or an ongoing reliability retainer.
Move platforms without moving your outage window
GitLab→Gitea, data-centre→cloud, VMs→Kubernetes, DMS/RDS database moves. I map the dependencies and blast radius first, cut over in reversible waves, and leave reusable IaC — a repository migration is a delivery-system migration. 700+ repos migrated to date.
Senior platform leadership, part-time
Embedded one to two days a week: golden paths, CI/CD and a GitOps operating model, cloud architecture, and coaching your engineers so the platform stops being one person’s tribal knowledge.
Not sure which fits? Book a free 20-minute call and we’ll scope it together.
Agencies & consultancies: I also deliver these white-label, under your paper.
ls case-studies/
Buyer-framed write-ups — the problem, what I owned, the decision, and how it was verified. The full field narratives live on OOMKilled.
A cloud-cost engagement run as an operating model, not a fire sale: right-sizing, Karpenter, and scheduled scale-down, with availability floors that were never up for negotiation.
A licensing squeeze forced a weekend cutover from GitLab to self-hosted Gitea. The hard part was never the clone — it was everything Git doesn't contain.
A botched rolling restart re-bootstrapped a fresh, empty database over two of three replicas. The only intact copy was a PVC the cluster had orphaned when it scaled down — and we should never have been that lucky.
tail -f oomkilled.com
I write up the incidents and migrations I live through at OOMKilled — war stories from inside real production, sanitised but not softened. It’s the same work I do for clients, told after the dust settles. If you want to know how I think before you hire me, start here.
Two engineers, ~700 repos, an on-prem GitLab, and a Monday deadline — and exactly how much hidden work a platform quietly does for you.
Read the dispatch →A botched rolling restart re-bootstrapped an empty database over two replicas. The only intact copy was a PVC Kubernetes had orphaned.
Read the dispatch →We put eBPF on the egress path of a fleet we were about to migrate. The gap between the docs and the network was the whole story.
Read the dispatch →ls ~/oss
Go CLI that scrubs public IPs and DNS from command output — safe log sharing for cluster debugging. Shipped via Homebrew.
AI DevOps extension for Azure DevOps — GPT-powered, context-aware pull-request feedback delivered inside the pipeline.
IaC for rapid, production-shaped Kubernetes bootstrapping on AWS with Terraform + Ansible — repeatable clusters, not hand-built ones.
Serverless image optimization on AWS Lambda + CloudFront — resize on demand at the edge, provisioned end-to-end with Terraform.
Production-shaped patterns in one POC: API gateway, async messaging (RabbitMQ), and distributed auth (JWT/RBAC) on Kubernetes.
End-to-end MLOps pipeline for medical imaging — a ResNet50 model served on AWS via FastAPI + Docker for real-time inference.
history | grep work
cat credentials
CKA and CKAD are performance-based, hands-on certifications from the CNCF & The Linux Foundation.
./contact-me
Tell me what’s on fire — a runaway bill, a shaky platform, a migration nobody wants to own. First call is free and we’ll figure out if I’m the right fit.
Named client references available on request.
