Deploy Multi-Cloud Infrastructure with Pulumi

This skill embodies a Pulumi infrastructure-as-code expert covering multi-cloud deployments, component architectures, and DevOps integration across AWS, Azure, GCP, and Kubernetes. Core resource-organization principles favor component resources for reusable, composable infrastructure units, consistent naming prefixes, relying on Pulumi's dependency graph for implicit ordering, and separating code into logical modules. State and configuration management emphasizes stack references for cross-stack dependencies, Pulumi's built-in secret encryption, and hierarchical configuration structured from a base layer down through environment-specific overrides.

A full component resource example builds a WebService class extending pulumi.ComponentResource, taking typed input args (name, image, optional port/replicas/environment) and producing a Kubernetes Deployment and Service, exposing the resulting load balancer URL as a typed pulumi.Output and registering it via registerOutputs. Stack configuration is shown both as a Pulumi.prod.yaml file (region, environment, instance type, an encrypted secret value, and a scaling object) and as typed TypeScript access via config.require, config.requireSecret, and config.requireObject with a generic type parameter.

A provider-agnostic Database component demonstrates multi-cloud abstraction: it branches on the stack name's provider prefix (for example aws-prod versus azure-dev) to construct either an AWS RDS instance or an Azure equivalent behind the same typed interface, assembling the connection string via pulumi.interpolate. Dynamic providers extend Pulumi to manage resources with no native provider by implementing create/update/delete methods that call an external REST API directly. Testing infrastructure code is shown using Mocha and Chai with pulumi.runtime.setMocks to stub resource creation, asserting on a component's generated Kubernetes spec without touching real cloud resources.

Use this when designing reusable Pulumi components, structuring multi-environment stack configuration, abstracting across cloud providers, writing custom dynamic providers for unsupported APIs, wiring CI/CD deployment pipelines, or enforcing infrastructure policy as code. A GitHub Actions workflow example runs on pushes to main touching an infrastructure path, installs dependencies, runs tests, previews the Pulumi stack, and deploys via the pulumi/actions GitHub Action with a PULUMI_ACCESS_TOKEN secret. A CrossGuard PolicyPack example enforces that S3 buckets have encryption configured, at mandatory enforcement level, failing a deployment that violates it.

A representative dynamic provider create handler:

const myProvider = new pulumi.dynamic.Provider("my-provider", {
    async create(inputs) {
        const response = await fetch(`https://api.example.com/resources`, {
            method: 'POST',
            body: JSON.stringify(inputs),
            headers: { 'Content-Type': 'application/json' }
        });
        const result = await response.json();
        return { id: result.id, outs: result };
    },

Inputs are typed component arguments and stack configuration values; outputs are typed pulumi.Output values representing deployed infrastructure state.

Covers Pulumi's providers for AWS, Azure, GCP, and Kubernetes, GitHub Actions via the pulumi/actions action for CI/CD, and the CrossGuard policy-as-code framework for automated compliance checks.

Platform and DevOps engineers building or maintaining Pulumi-based infrastructure who need concrete patterns for reusable components, multi-cloud abstraction, dynamic providers, automated testing, CI/CD deployment, and policy enforcement.