Five years building and running infrastructure — CI/CD, Kubernetes, Terraform, and bare-metal virtualisation.
I build and run the infrastructure other teams ship on.
DevOps · SRE
Platform · Cloud
I build and run the infrastructure other teams ship on. Five years across DevOps, SRE, platform and cloud — pipelines, Kubernetes, Terraform, and bare-metal virtualisation. Most of it in Go and Linux, close to the kernel and close to the network. What I actually optimise for is boot time, pipeline time, and a quiet pager.
Four roles, moving steadily from writing services toward running the platforms they live on. Every number below is something I measured myself, in production.
Grouped by what I actually do with them rather than by tool name — the tools are in the next section.
Go services under real concurrency — gRPC, SSE, event-driven work on Kafka and RabbitMQ, and multi-tenant data models that stay fast as tenants pile up.
Prometheus, Grafana, Loki and OpenTelemetry wired in from the start. SLOs that mean something, alerts that fire for a reason, and post-incident notes people actually read.
Jenkins and GitHub Actions driving Terraform, Ansible and Packer. Every environment reproducible from a repo, every change reviewable before it lands.
QEMU/KVM and QMP, cloud-init, PXE, HAProxy and nftables. Comfortable below the container layer, where most people stop looking.
Where I'm not strong, so you don't find out later: Java and Spring Boot are intermediate for me and I have never used Hibernate. My English is intermediate — fine in writing and in a technical conversation, still catching up in a fast meeting.
Split honestly: the first row is what I touch most weeks, the second is what I reach for when a project needs it.
Any of these four, at a team that runs its own infrastructure rather than outsourcing the interesting parts.
Pipelines, infrastructure as code, and the delivery path teams use every day.
SLOs, observability, capacity and incident response for systems that have to stay up.
Internal platforms and self-service infrastructure, so product teams stop filing tickets.
AWS and hybrid networking, with cost and security treated as design constraints.
One month notice. Happy to talk through any of the numbers on this page in detail — or ask the character up top.