Internal Developer Platform


Difficulty	Advanced
Team Size	3-5 people
Time	~35-45 hours
Demo-ready by	Step 5
Prerequisites	Node.js, React or similar, API design, DevOps concepts
Built by	Backstage (Spotify), Port, Cortex, OpsLevel

Skills you'll earn: Service catalog design, template scaffolding, plugin architecture, API aggregation, developer portal UX

Start by listing services from a catalog file. End with a Backstage-like portal where teams discover services, provision infrastructure, and use templates to bootstrap new projects.

Step 1: Build a service catalog (~2-3 hours)

Developers can't use what they can't find.

Define services in YAML files: name, owner, description, repo URL, docs URL, API endpoint
Build an API that reads and serves the catalog: GET /services, GET /services/:name
Use Express or Fastify

You now have: A service directory.

Step 2: Web dashboard (~3-4 hours)

A JSON API is for machines. Developers want a UI.

Build a web UI listing all services with search and filter
Service detail page: description, owner, links to repo/docs/API, health status
Group services by team or domain
Use React or Vue

You now have: A developer portal.

Step 3: Health and dependency visualization (~4-5 hours)

You know what services exist. You don't know how they connect or if they're healthy.

Each service declares its dependencies in the catalog YAML
Render a dependency graph on the dashboard (use D3.js or React Flow)
Poll health endpoints and show status badges (green/yellow/red)
Click a service to see its upstream and downstream dependencies

You now have: Service health and topology view.

Step 4: Project scaffolding templates (~4-5 hours)

Starting a new service from scratch is slow and inconsistent.

Define templates: "Node.js API", "React Frontend", "Python Worker"
Each template is a directory of files with placeholder variables (, )
API endpoint: POST /scaffold with template name and variables
Generate the project, push to a new git repo, register in the catalog

You now have: One-click project bootstrapping.

Step 5: Infrastructure provisioning (~3-4 hours)

A new service needs a database, a queue, a DNS entry.

Define resource types: PostgreSQL database, Redis cache, S3 bucket
Provisioning API: POST /provision with resource type and config
Backend calls Terraform or Docker Compose to create the resource
Return connection details and inject them into the service's config

You now have: Self-service infrastructure.

Step 6: Documentation hub (~4-5 hours)

Aggregate and render docs from each service's repo (README, OpenAPI specs)
Auto-generate API documentation from OpenAPI specs
Search across all service documentation

Step 7: RBAC and audit (~3-4 hours)

Team-based access control: only service owners can modify their catalog entry
Audit log of all provisioning and scaffolding actions
Integration with SSO (OAuth2/OIDC)

Useful Resources

Where to go from here

Cost tracking per service (aggregate cloud bills by service owner)
Incident management integration (PagerDuty, Opsgenie links per service)
Scorecards (rate services on documentation quality, test coverage, security posture)
Plugin system (let teams add custom tabs to service pages)
ChatOps integration (provision resources from Slack)

Internal Developer Platform ​

Step 1: Build a service catalog (~2-3 hours) ​

Step 2: Web dashboard (~3-4 hours) ​

Step 3: Health and dependency visualization (~4-5 hours) ​

Step 4: Project scaffolding templates (~4-5 hours) ​

Step 5: Infrastructure provisioning (~3-4 hours) ​

Step 6: Documentation hub (~4-5 hours) ​

Step 7: RBAC and audit (~3-4 hours) ​

Useful Resources ​

Where to go from here ​