Production infrastructure R&D environment

Outcome

• Implemented resilient storage and clear recovery paths, as covered in The Platform Mindset.
• Applied zero-trust networking with segmented VLANs and least-privilege access, further detailed in Architecting Services.
• Operationalised GitOps, monitoring, and incident-ready runbooks, with lessons captured in Heroics.

Operating targets

• 99.9% availability target for core services, informed by the resiliency review in Year-end notes: homelab resolutions and the backlog I am finally tackling.
• RTO: 2 hours, RPO: 30 minutes for critical data sets.
• 20+ self-hosted services with capacity headroom for growth.

Focus areas

• Platform architecture and capacity planning for predictable performance
• Security-first access controls, backup hygiene, and change control
• Automation that reduces risk and makes changes auditable
• CI/CD pipelines for repeatable deployments and safe rollbacks, plus runner isolation details in GitLab CI/CD Runners

What I owned

I own the platform end-to-end, with production-level standards for reliability, security, and operational discipline that mirror how a lean internal team would run it.

Source

All infrastructure code, CI/CD pipelines, and configuration live on a self-hosted GitLab CE instance. The platform runs on the same hardware it deploys to: GitLab CE, its Docker-in-Docker runner, and every service it ships all sit on the same Proxmox host. This is deliberate: it means the platform is self-sustaining and doesn't depend on any external SaaS for its own delivery pipeline.