SRE

We wired AI into real delivery pipelines: PR review, agentic end-to-end testing, security testing in CI, LLM observability, and alert triage. Here is what helped, what it saved, and where it quietly lied to us.

A plain definition of the fractional platform engineer model: senior platform and SRE capacity on a fraction of a week, what it is not, when it fits a 10 to 50 person engineering org, and what a typical month looks like.

A practical checklist for reading the health of a production Kubernetes cluster: pod pressure, restart causes, the requests-versus-usage gap, HPA and PDB sanity, certs, registry health, control plane, quotas, and alert fatigue.

Horizon connects sprint delivery, Kubernetes runtime, and AI-generated analysis in one platform. Here is the problem it solves and where it is today.