Stage 7 · Master
Platform Engineering Fundamentals
Platform Maturity Model
Assess your platform maturity from Level 0 (chaos) to Level 4 (intelligent) and plan incremental improvements.
7 min readPlatform Engineering & Developer ExperienceMaster
Five Maturity Levels
The Platform Maturity Model helps you understand where you are and what to invest in next. Most organizations start at Level 0-1 and progress incrementally. Skipping levels leads to fragile platforms.
| Level | Name | Characteristics | Platform Team Focus |
|---|---|---|---|
| 0 | Chaos | Manual everything, ticket-driven, no standards | Firefighting, keeping lights on |
| 1 | Scripted | Automation scripts, some CI/CD, wiki docs | Maintaining scripts, reducing toil |
| 2 | Self-Service | Internal CLI, basic portal, golden paths | Building APIs, templates, self-service |
| 3 | Productized | Portal with catalog, PM, SLIs, feedback loops | Product management, DX, platform SLIs |
| 4 | Intelligent | AI-assisted, predictive, auto-remediation | ML platform, anomaly detection, auto-optimization |
Self-Assessment Framework
Rate each dimension 0-4. Your overall level is the lowest dimension score (weakest link). Focus investment on raising the floor.
| Dimension | Level 0 | Level 1 | Level 2 | Level 3 | Level 4 |
|---|---|---|---|---|---|
| Interface | Tickets only | Scripts + wiki | CLI + basic portal | Full portal + IDE | AI chat + predictive |
| Scaffolding | Copy-paste repo | Cookiecutter/Yeoman | Parameterized templates | Scaffolder + policy | AI-generated from spec |
| CI/CD | Manual Jenkins | Shared pipeline lib | Golden path pipelines | Policy-enforced, tested | Auto-optimizing pipelines |
| Observability | None / ad-hoc | Basic dashboards | Defaults in template | SLOs, auto-dashboards | Anomaly detection, RCA |
| Security | Manual reviews | Checklists in CI | Policy as code (admission) | Supply chain security | Auto-remediation |
| Environment | Manual VMs | Terraform modules | Ephemeral PR envs | Self-service + TTL | Predictive scaling |
| Governance | Tribal knowledge | Wiki + reviews | Policy engine + catalog | Scorecards + compliance | Continuous compliance |
| Feedback | None | Annual survey | In-app + quarterly | Real-time + NPS | Predictive friction detection |
| Team | No platform team | Part-time / shared | Dedicated platform team | PM + UX + Eng | ML engineers + data scientists |
Level 0: Chaos
- Every team manages their own infrastructure
- Deployments via SSH, manual kubectl, or custom scripts
- No standard CI/CD — each team builds their own
- Observability: 'check logs on the server'
- Security: manual reviews, inconsistent
- Onboarding: weeks to first deploy
- Incidents: hero mode, no runbooks
Don't Stay Here
Level 0 is the default starting state. The goal is to move to Level 1 quickly by identifying the most painful manual processes and scripting them.
Level 1: Scripted
- Automation scripts for common tasks (provision DB, create repo, deploy)
- Shared CI/CD pipeline library (e.g., shared Jenkins library, GitHub Actions reusable workflows)
- Documentation in wiki/Confluence — but often outdated
- Some golden path attempts — but not enforced or maintained
- Platform work done by 'platform-interested' engineers part-time
- Still ticket-driven for anything non-standard
Level 2: Self-Service
- Dedicated platform team (3+ engineers)
- Internal CLI (
platform service create,platform db provision) - Developer portal (Backstage or custom) with software catalog
- Golden path templates with scaffolder (Backstage Scaffolder, Cookiecutter + CI)
- Policy as code at scaffold time + admission control (OPA/Kyverno)
- Ephemeral preview environments for PRs
- Platform SLIs tracked (scaffold success, API latency)
- Developer feedback: in-app widgets, quarterly survey
Level 3: Productized
- Platform PM (dedicated or strong tech lead acting as PM)
- UX designer for portal/CLI
- Platform roadmap driven by developer research, not stakeholder requests
- Versioned golden paths with 18-month support, automated migration tooling
- Platform SLIs with error budgets, on-call for platform APIs
- Developer NPS tracked quarterly, action plans published
- Enabling teams (Security, SRE, Data) contributing to platform
- Cost visibility per team/service in portal (showback/chargeback)
- Compliance automation: policy reports, audit evidence generation
Level 4: Intelligent
- AI-assisted scaffolding: 'Create a payment service with PostgreSQL and Kafka' → generates complete repo
- Predictive DX: ML models predict CI failures, suggest fixes, auto-generate runbooks from incidents
- Auto-remediation: Platform detects drift, scales, rotates certs, patches base images without human intervention
- Intelligent routing: Platform routes traffic based on cost, latency, carbon footprint
- Natural language platform interface: ChatOps + LLM for platform operations
- Platform team includes ML engineers, data scientists
- Continuous optimization: Rightsizing, spot instance orchestration, carbon-aware scheduling
Progression Strategy
- Assess: Run the self-assessment across all dimensions
- Identify Floor: Your level = lowest dimension score
- Pick One Dimension: Focus on raising the floor (e.g., if Security is Level 1, invest in Policy as Code)
- Quick Wins: Script the most painful manual process (Level 0→1)
- Build Platform Team: Hire/assign 2-3 engineers full-time (Level 1→2)
- Launch Portal + Scaffolder: Backstage or custom (Level 2 milestone)
- Hire PM: When platform team >5 engineers (Level 2→3)
- Invest in DX: UX, feedback loops, NPS, SLIs (Level 3)
- Experiment with AI: Copilot for platform, predictive scaling (Level 3→4)
Maturity Is Not Linear
You can be Level 3 in Scaffolding but Level 1 in Security. That's normal. The model helps you see the gaps. Don't try to be Level 4 everywhere — be Level 3 in your core capabilities, Level 2 in emerging ones, and have a plan for the rest.
Mark this lesson complete to store local progress and unlock a cleaner resume path the next time you visit.