DevOps & platform engineering
We design, build and run cloud infrastructure for product companies: AWS and Kubernetes, GPU capacity for AI workloads, delivery pipelines, cost, security and compliance, observability. More than fifteen years of hands-on work, remote from Bucharest, for teams across the UK and EU.
The work a platform team would do, delivered by someone who has run it in production, been on call for it, and led the team doing it.
EKS, AKS, OKE and other managed clusters, or self-managed ones with Rancher. Upgrades without downtime, node pools and autoscaling with Karpenter or Cluster Autoscaler, Helm and GitOps, fleet management with tools like Rafay.
GitHub Actions, Azure DevOps or GitLab, self-hosted runners where they pay off, caches that make builds fast, deploy pipelines with gates and rollbacks, and an end to flaky end-to-end suites.
Spend audits with attribution per service, savings plans and right-sizing, NAT and egress analysis, and removal of what nobody uses. Findings first, then the changes.
The infrastructure side of ISO 27001 and Cyber Essentials with Vanta or Drata, CDN and WAF in front of public endpoints, secrets in Vault, image and CVE hygiene, and answers for customer security questionnaires.
Prometheus, Grafana and Loki set up properly, alerts tuned so they page only when it matters, incident playbooks, and the fixes that stop the same incident coming back.
GPU-enabled EKS with NVIDIA tooling and scale-from-zero node pools, self-hosted vLLM inference, LLM agents as production workloads, and coding agents that operate CI and infrastructure under review.
Three ways to engage. All of them end with your team able to run what we built.
A fixed number of days a month inside your team: your Slack, your repos, your incidents. We own the platform backlog so your developers can keep shipping.
An EKS upgrade, a CI migration, a cost audit, an ISO 27001 push. Scoped and priced up front, delivered with the runbooks to keep it running.
A close look at your architecture, pipelines or cloud bill, with a written report and a prioritised list of what to change first.
How a stale Hugo blog became a firm site in one evening with Claude Code: what the agent did, the pain points it found and fixed, the GitHub Pages deploy that …
Notes from running an LLM agent as a production workload on Kubernetes and letting coding agents operate the platform: what stayed the same, what changed, and …
How we moved a WordPress site off AWS onto Hugo and GitHub Pages without visitors noticing — capture, verify byte-for-byte, cut over DNS, and stop paying for an …
Tell us what is slow, expensive or fragile. We reply within one business day.
Get in touch