CI/CD Pipelines
Automate build, test and deployment so every change reaches production safely and quickly.
Continuous integration (CI) builds and tests every change as soon as it is pushed. Continuous delivery (CD) packages each passing change so it can be released at any time; continuous deployment releases it automatically.
A good pipeline is fast, repeatable and the only way code reaches production.
Diagram
Key concepts
- Build once, deploy many
- The same artifact (container image or package) moves from staging to production, so what you tested is what you ship.
- Pipeline as code
- Pipelines live in the repo as YAML (GitHub Actions, Azure Pipelines, GitLab CI).
- Quality gates
- Tests, coverage, security scans and approvals that must pass before promotion.
- Environments
- Dev, staging and production with their own configuration and secrets.
- Infrastructure as code
- Bicep, Terraform or CloudFormation deployed by the same pipeline.
Trade-offs
Benefits
- Small, frequent, low-risk releases
- Bugs found minutes after they are written
- Repeatable deployments with an audit trail
Costs
- Pipelines need maintenance and fast test suites
- Flaky tests erode trust quickly
Use it when
- Every team, from day one of a project
Common mistakes
- Rebuilding the artifact per environment
- Secrets in pipeline YAML instead of a vault or OIDC federation
- Manual production steps that bypass the pipeline
- Pipelines that take an hour, so people batch changes
Interview questions
Q1Continuous delivery vs continuous deployment?
In continuous delivery every change is releasable but a person decides when to release. In continuous deployment every passing change goes to production automatically.
Q2How do you keep pipelines fast?
Cache dependencies, run test suites in parallel, run slow end-to-end tests only on main or nightly, and keep the unit-test layer large and quick.
Q3How should a pipeline authenticate to the cloud?
With workload identity federation (OIDC) and short-lived tokens rather than long-lived secrets stored in the CI system.
Related lessons
Cloud & Delivery
Deployment Strategies
Rolling, blue-green, canary and feature flags — release without downtime and roll back fast.
Cloud & DeliveryHow Git Works
Working directory, staging area, local and remote repositories — and the commands that move changes between them.
Cloud & DeliveryKubernetes Architecture
The control plane, worker nodes and core objects that run containers at scale.