Each follows the [verb] [object] [number] structure hiring managers grade against. Copy them as a starting point, swap in your own numbers, and read the annotation to understand why each one works.
Paved road
Built the paved-road service scaffolder (Backstage Software Templates + Crossplane compositions); 84% of new services since launch (28 of 33) shipped on the paved road with no platform-team-side work.
Why it works: Names the tooling, adoption ratio with absolute count, and the zero-touch property. Platform-engineering bread-and-butter.
Time to deploy
Cut time-to-first-deploy for new services from 9 days (avg) to 2 hours (p95) via the Backstage Software Template + Crossplane composition + ArgoCD ApplicationSet pipeline.
Why it works: Names the metric (with avg→p95 senior calibration), the before/after, and three pipeline components.
DORA
Lifted the org-wide DORA metrics: deploy frequency 1.2×/week → 4×/day; lead time 6 days → 14 hours; change-failure rate 18% → 6.2%; MTTR 4h → 38m.
Why it works: Names all four DORA metrics with before/after. The complete set signals platform-engineering depth.
Self-service
Shipped 12 self-service capabilities through 2024 (env provisioning, on-call setup, observability defaults, secret rotation, RDS, EKS namespace, CI/CD, service scaffold, ArgoCD onboarding, IAM, DNS, log aggregation); avg 78% reduction in platform-team tickets per capability.
Why it works: Names the capability count, lists examples, and quantifies the ticket-deflection outcome.
Delivery
Migrated the org from Jenkins + custom scripts to ArgoCD ApplicationSets; 38 services onboarded in 8 weeks; deploy failure-rate-attributable-to-platform fell from 11% to under 0.4%.
Why it works: Names the migration (old → new), service count, duration, and the platform-attributable failure-rate outcome. The 'attributable' detail is senior calibration.
Policy
Authored the team's policy-as-code library (Kyverno + OPA); 38 policies enforced cluster-wide (image signing, label requirements, resource limits, network policies); zero policy-violation incidents over 6 months.
Why it works: Names the tools (Kyverno + OPA), the policy count, the categories, and the zero-incident window.
Backstage
Built the developer-portal experience in Backstage with the Software Catalog + TechDocs + Cost Insights + Lighthouse plugins; weekly active developer users in the portal reached 240 of 280 (86%) within 4 months of launch.
Why it works: Names the plugins, the WAU metric, and the adoption ratio. Backstage adoption with a WAU number is platform-engineering-specific senior signal.
Security
Built the team's golden-path runtime image library (Wolfi-based, distroless, vulnerability-scanned in CI); 92% of services migrated within 6 months. Reduced CVE backlog from 380 open to 14.
Why it works: Names the image-library architecture (Wolfi + distroless), the migration adoption, and the CVE outcome with absolute numbers.
On-call
Authored the on-call paved-road (PagerDuty + opsgenie-compatible templates + auto-generated runbooks); new service on-call setup time fell from 3 days to 12 minutes.
Why it works: Names the tooling and the setup-time outcome. On-call provisioning is a platform-engineering responsibility.
Open Source
Two merged PRs to backstage/backstage core — one closed a CatalogApi pagination bug under entity counts above 5,000; one extended the GitHub Auth provider with org-based scope enforcement.
Why it works: Named project (Backstage), two PRs, technical descriptions that signal depth (5k entity pagination, org-based scope).
Plugin authorship
Published the backstage-plugin-finops-aws plugin to the marketplace; 38 weekly active downloads at 6 months; cited in two external platform-engineering blog posts.
Why it works: Named plugin, marketplace presence, download metric, external citations. Backstage plugin authorship is the platform-engineering open-source currency.
Mentorship
Mentored 2 platform engineers from generalist SWE into platform-engineering focus; both shipped sole-owner self-service capabilities within 6 months.
Why it works: Names the mentee trajectory, the timeframe, and the deliverable. Platform-engineering mentorship is hard to demonstrate without an outcome.