Reference

Platform engineering glossary

Thirty terms every engineering leader should know, with plain-English definitions. No vendor references. Updated for 2026.

Platform engineering has accumulated its own vocabulary over the last decade. Some terms are rigorous (DORA metrics, SPACE framework); others are folk (YAML engineer, golden path) but still useful. Below is the working vocabulary for engineering leaders at organisations considering, starting, or scaling a platform function.

A C D F G I K O P S T V Y Z

Adoption rate: Share of product engineers actually using the platform's golden paths. The single most important platform success metric; aim for over 70% within 18 months.

CI/CD: Continuous integration and continuous delivery. Automated build, test, and deploy pipelines. The most mature category of platform tooling.
Cognitive load: The amount of context an engineer must hold to do a task. Platform engineering exists to reduce cognitive load on product teams.
Container orchestration: Managing containerised workloads at scale. Kubernetes is the dominant primitive; platform teams usually wrap it with higher-level abstractions.

DevEx: Developer experience. A subset of platform engineering focused on the developer's daily friction: editor tooling, local dev environment, onboarding, documentation.
DORA metrics: Four operational metrics from the DevOps Research and Assessment programme: deployment frequency, lead time, change failure rate, failed deployment recovery.

Fleet management: Coordinated configuration and upgrade of a large number of runtime instances. Common for serverless functions, edge workers, agents.

Golden path: An opinionated, paved route through the platform. Product teams who follow it get observability, deployment, and security for free.
GitOps: Infrastructure deployed from git state rather than imperative commands. Gives an audit trail and natural rollback.

IDP (Internal Developer Platform): The product platform engineering builds. Not a specific tool; an assembly of capabilities that together give product teams self-service access.
Infrastructure-as-code: Cloud resources defined in versioned declarative files rather than clicked in a console. Foundational for platform engineering.

Kubernetes operator: A custom controller that extends Kubernetes with domain-specific logic. Platform teams write them to encode operational patterns.

Observability: The ability to understand an internal system's state from its external outputs. Logs, metrics, traces combined.
On-call rotation: Scheduled responsibility for responding to production incidents outside business hours. Carries stipend and productivity overhead.

Platform as a product: Treating the internal platform as a product with users (product engineers), adoption metrics, and a roadmap driven by user research.
Platform engineer: An engineer who builds and runs the internal developer platform. Part infrastructure, part product, part developer experience.
Platform team: The team responsible for the IDP. Sized roughly 5-10% of total engineering.
Policy as code: Organisation policies (security, compliance, cost) expressed as programmable rules evaluated automatically in pipelines.
Pull-based delivery: An agent in the target environment pulls desired state, rather than CI pushing to it. Common in GitOps.

Secrets management: Central storage, rotation, and auditing of credentials that services need to function. A core platform capability.
Self-service infrastructure: Product engineers provisioning what they need without a ticket to the platform team. The primary platform goal.
Service catalogue: Registry of every service in the organisation with owners, documentation, runbooks, and scorecards. The starting point for a developer portal.
Service level objective (SLO): A target for reliability of a service, expressed as a percentage over a time window. Platform teams usually set and report SLOs.
Site Reliability Engineering: A software engineering approach to operations. Shares 70-80% of its practice with platform engineering; the labels are often used interchangeably.
SPACE framework: Productivity measurement framework: Satisfaction, Performance, Activity, Communication, Efficiency. A useful supplement to DORA.

Technical debt: Accumulated cost of past shortcuts. Platform teams both reduce product-engineering debt and accumulate their own platform debt.
Time to productivity: The wall-clock time from a new engineer's first day to their first shipped change. A proxy for platform quality.

Value stream: The sequence of steps from idea to shipped software. Platform engineering optimises the value stream; DORA metrics measure it.

YAML engineer: Dismissive term for a platform engineer who spends their days configuring declarative tools rather than writing code. Real practitioners are both.

Zero-trust architecture: Security model where no request is trusted by network location alone. Every service-to-service call is authenticated and authorised.

Looking for more depth?

Most of these terms connect to a deeper page on this site. The team structure page covers roles and ratios. The ROI page goes deeper on DORA and DX Core 4. The tooling budget page covers CI/CD, observability, secrets, and service catalogue as budget categories. The hidden costs page covers on-call, documentation, and advocacy.