Infrastructure as code
Terraform and reusable modules with remote state, locking, and a reviewed plan in every merge request. No manual console changes.
Infrastructure · DevOps · Homelab
We design and operate infrastructure that holds up in production
Forgekloud builds platforms on Kubernetes, Terraform, and AWS, then documents the real decisions: what worked, what it cost, and what we would do differently next time.
terraform apply -auto-approve
Plan: 14 to add, 3 to change, 0 to destroy.
Apply complete. Resources: 14 added, 3 changed.
kubectl get nodes -o wide
forge-cp-1 Ready control-plane 132d v1.31.4
forge-w-1 Ready <none> 132d v1.31.4
forge-w-2 Ready <none> 132d v1.31.4
curl -s https://forgekloud.dev/healthz
{"status":"ok","uptime":"132d 04h"}
00
Published projects
With architecture and decisions
00
Technologies covered
From the real stack, not a CV
99.9%
Homelab uptime
Last 90 days
< 5 min
Commit to production
Complete pipeline
What we build
Three fronts, one discipline
Everything here is reproducible: if it cannot be built from scratch with one command, it is not finished.
Kubernetes in production
Clusters with GitOps, network policies by default, real resource limits, and a tested cold-start strategy rather than an assumed one.
Observability and delivery
Pipelines that fail early and metrics that answer specific questions: what broke, since when, and who is affected.
This site is infrastructure too
Node and SQLite in an unprivileged container behind a reverse proxy with automatic TLS. The publishing panel uses signed sessions, CSRF, and byte-validated image uploads.
View the technical colophonRecent work
Projects
Every entry includes the problem, architecture, decisions, and what went wrong.
No posts yet
Do you have a platform that fails on Fridays?
Write to us with the context and we will tell you what we would tackle first. No four-step forms.