AWS DevOps Engineer

Tests that run only on a developer laptop protect nobody. You place unit, integration, security and load tests where they give fast feedback, and make failing results block promotion.

• Unit and integration tests in build stages
• Static analysis and dependency scanning
• Ephemeral test environments
• Test reports and quality gates
• Load and smoke tests after deployment
• Keeping the pipeline fast as tests grow

You build A pipeline with test reports and quality gates that stop a bad build.

How you version and deploy an artifact decides how safely you can undo it. You manage container images and packages, then choose rolling, blue/green or canary releases for instances, containers and Lambda functions.

• Versioning and retention in ECR and CodeArtifact
• Immutable artifacts and promotion by reference
• Rolling, blue/green and canary releases compared
• CodeDeploy for EC2, ECS and Lambda
• Traffic shifting and automatic rollback
• Database changes alongside application releases

You build A canary release with automatic rollback on a failing alarm.

Infrastructure you cannot recreate is infrastructure you cannot trust. You write reusable templates, review changes before they apply, and handle drift, dependencies and custom logic in CloudFormation and CDK.

• Template structure, parameters and outputs
• Nested stacks and reusable modules
• Change sets and safe updates
• CDK constructs and testing infrastructure code
• Drift detection and resource import
• Custom resources and stack policies

You build A reusable infrastructure module tested and deployed through a pipeline.

Real organizations run dozens of accounts. You automate account creation, baseline security and stack rollout across accounts and Regions, so every environment starts the same way.

• Account vending with Organizations and Control Tower
• StackSets for baselines across Regions
• Service Catalog for approved building blocks
• Cross-account roles for pipelines
• Systems Manager across accounts
• Handling Regional differences and quotas

You build An account baseline that rolls out to three accounts in two Regions.

Repetitive operations should be code. You write scripts against the AWS SDK and CLI, run them from Lambda, Step Functions and Systems Manager, and make them safe to run twice.

• SDK and CLI patterns for fleet operations
• Idempotent scripts and error handling
• Lambda for operational automation
• Step Functions for multi-step runbooks
• Systems Manager Automation and Run Command
• Scheduling with EventBridge

You build An automated runbook that patches a fleet in stages and reports the result.

Availability is designed in layers. You spread workloads across zones, scale on the right signals, and shape load balancing, DNS and data replication so that one failure does not take the service down.

• Auto Scaling groups and scaling policies
• Load balancers and health checks
• Container and serverless scaling
• Multi-AZ and multi-Region data replication
• DNS routing policies for availability
• Load testing to find the real limit

You build A multi-AZ service that scales on load and survives a zone failure.

Recovery objectives are promises you must be able to keep. You choose a disaster recovery pattern, automate backup and failover, and rehearse recovery so the time and data loss you claim are the ones you measure.

• Setting RTO and RPO per workload
• Backup automation with AWS Backup
• Pilot light, warm standby and active-active
• Automating failover and failback
• Fault injection and game days
• Verifying restores

You build An automated failover to a second Region with a measured recovery time.

You cannot fix what you cannot see. You collect application, system and AWS logs and metrics centrally, shape them into useful queries, and keep them across accounts without losing control of cost.

• CloudWatch agent, metrics and custom metrics
• Structured logging and Logs Insights queries
• Log aggregation across accounts and Regions
• Tracing with X-Ray and OpenTelemetry
• Retention, archiving and cost
• Dashboards for service owners

You build A centralized logging and metrics setup across two accounts with shared dashboards.

Good alerts are rare and specific. You build alarms from metrics and logs, reduce noise with composite and anomaly-based alarms, and route events to the people and automation that should act.

• Metric alarms and thresholds that mean something
• Composite alarms and anomaly detection
• Synthetic checks for user-facing paths
• EventBridge event patterns and routing
• Notification and escalation paths
• Reviewing alerts that fired for no reason

You build An alarm and event routing design that pages only on user impact.

Many failures have a known fix that a machine can apply. You trigger remediation from events and rules, and learn to read failed deployments, stuck stacks and unhealthy tasks to find the real cause.

• Remediation with EventBridge, Lambda and Systems Manager
• Config rules with automatic remediation
• Diagnosing failed CloudFormation and CodeDeploy runs
• Unhealthy instances, tasks and pods
• Reading Lambda and API errors
• Post-incident review and follow-up automation

You build A rule that detects a misconfiguration and fixes it automatically, with an audit trail.

Security checks that depend on people fall behind. You manage access at scale, keep secrets out of code, run continuous detection and scanning, and produce audit evidence automatically.

• IAM roles and least privilege for pipelines
• Secrets Manager and Parameter Store in deployments
• Encryption defaults and key access
• GuardDuty, Inspector and Security Hub in the pipeline
• Config and CloudTrail for auditing
• Policy as code and guardrails

You build A pipeline with secret handling, scanning and an audit report.

Combine every part into one platform. You deliver a service across accounts and Regions through a tested pipeline, watch it with meaningful alerts, and prove it recovers on its own when you break it.

• Designing the platform end to end
• Building the pipeline and infrastructure code
• Wiring monitoring and remediation
• Breaking it on purpose and measuring recovery
• Presenting the platform and its runbooks

You build A multi-Region delivery platform with pipelines, alarms, remediation and a recovery test.