Infrastructure

I try it at home before production

Most of what I run in production I first set up, broke and fixed on my own hardware. That is how I can run the platform at work without a separate DevOps team.

Why I bother

I got into this industry by self-hosting things, running game servers and writing bots, and I never stopped. It is still the quickest way I know to find out how a piece of infrastructure really behaves.

Home lab and self-hosting

Raspberry Pi clusters and home servers running as a permanent lab: GitLab runners, sandbox and demo environments for my own products, and staging for new architectures before they reach production.
A power-management layer built from Raspberry Pi Zeros and Wake-on-LAN, so servers come up when they are needed instead of running all day.
A self-hosted game-server control panel, and a set of chat bots I wrote and run on my own servers.

Security, provisioning and virtualisation

I set up Linux servers from scratch, at home and in production, including reverse proxies, TLS termination and network segmentation.
I design and run Proxmox environments and LXC containers for both lab and company workloads.
Provisioning and security hardening are automated with Ansible, so environments can be rebuilt instead of hand-tuned.

Observability and platform tooling

Grafana dashboards on Prometheus for metrics and performance, in production and across the home setup.
Distributed tracing with Jaeger and .NET Aspire, so latency and failures between services can be seen instead of guessed.
Milvus for vector storage behind AI and retrieval workloads.

Networking and edge

At work I run the edge: DNS, routing, WAF and security policy. I run the same for my own domains, plus firewalls, VPN access, tunnels and DNS across a mix of home and cloud.

Learning now

ArgoCD: Deployments are moving to GitOps, already done for my own products and in progress at work.
Terraform: Declarative infrastructure for the pieces that are still provisioned by hand.

Where it paid off

Where the lab made a difference at work.

Every infrastructure migration I have led at work (IIS to Docker, Docker to Kubernetes, monolith to event-driven) ran on this hardware before it touched anything a client could see.

Container orchestration, reverse proxying and observability all reached production through the lab first, which is why the platform at work runs without a separate DevOps function.

My own products get a real staging environment from the same setup, so Notevex, TableButler and Operavex ship against infrastructure shaped like production, not a laptop.

Let's build
something.

I am happy to walk you through any system on this page, or go through its architecture in depth. Email or LinkedIn is quickest.

[email protected]LinkedInDownload CV

I try it at home before productionItryitathomebeforeproduction