Context
Linewise's nine squads shared one monorepo with a 40-minute CI run and a deploy queue coordinated in a Slack channel. A single SRE team owned every production change.
Deploys averaged two per day for the whole company — and incident response meant log archaeology across six tools.
The challenge
Squads were blocked on each other and on SRE; ownership boundaries didn't exist in the infrastructure even though they existed on the org chart.
MTTR was measured in hours because finding the right logs took longer than fixing the bug.
Approach
Paved roads
Weeks 1–6Backstage developer portal with golden-path templates: new service, new queue worker, new cron — each scaffolds repo, pipeline, dashboards, and ownership metadata in one step.
CI/CD overhaul
Weeks 4–10Reusable GitHub Actions workflows with test sharding and a merge queue cut CI from 40 to 9 minutes; per-squad deploy pipelines removed the shared queue entirely.
Unified observability
Weeks 8–12Grafana + Loki consolidated six tools into one. Every golden-path service ships with dashboards, SLOs, and alert routing to its owning squad.
You-build-you-run enablement
Weeks 12–14On-call training, incident-review process, and scorecards in Backstage. SRE moved from gatekeeper to platform team.
Architecture
Results
Up from 2/day company-wide; squads deploy independently.
Down from hours — alerts route to owners with the right dashboard linked.
Down from 40 minutes via sharding and the merge queue.
100% adoption within one quarter, no mandate required.
“The platform sells itself: the golden path is genuinely the easiest way to ship. Nobody has opened the old Slack deploy channel in months.”