AWS vs Azure vs GCP Skills Tests: Cloud Platform Assessment Guide

The cloud platform hiring problem

You need a senior DevOps engineer. You have five great candidates. Two are deep AWS; one is Azure-strong; one uses GCP; one is cloud-agnostic but less operational depth.

Who do you hire? It depends. But most teams get this wrong: they either hire for the specific platform and create lock-in risk, or they hire the generalist and discover the platform knowledge gap is deeper than expected.

The right answer is to assess both: platform-specific operational knowledge and transferable systems thinking.

When to test platform-specific skills

Test for AWS/Azure/GCP depth if:

• Your existing stack is locked into that platform
• Your hiring need is to deepen expertise in that cloud
• Your contract or compliance requirements necessitate a specific cloud

Test for cloud-agnostic fundamentals if:

• You're multi-cloud or might migrate
• You're hiring for systems thinking over tool mastery
• The candidate will work across platforms

Most teams should test both: "Can they operate AWS well?" AND "Can they reason about architecture platform-independently?"

AWS assessment framework

Skills to test

1. IAM and security modeling

   • Can they design least-privilege access for a microservices architecture?
   • Do they understand SCPs vs. inline policies vs. managed policies?
   • Can they explain why you shouldn't use root credentials?

2. Networking and connectivity

   • VPC design: subnets, routing, NACLs, security groups
   • When to use direct connect vs. VPN vs. Internet Gateway
   • Cross-account or cross-region networking trade-offs

3. Storage and databases

   • EBS optimization: gp3 vs. io2, provisioned throughput
   • RDS failover and backup strategies
   • S3 lifecycle policies and versioning for data retention

4. Compute scaling

   • EC2 Auto Scaling Groups vs. Spot Instances
   • ECS vs. EKS trade-offs
   • Lambda cold start implications for production workloads

Assessment question examples

Take-home scenario:

"Design a system to ingest 10GB of logs per day from 100 EC2 instances, store them durably, and query them with sub-second latency. You have a budget of $2,000/month. AWS only. Explain your choices."

Good answer considers:

• CloudWatch Logs for collection (managed) vs. Firehose + S3 + Athena (cost optimized)
• Log retention policies to manage costs
• Indexing strategy (Athena partitioning, or Elasticsearch)
• Query patterns and whether sub-second latency is real or assumed

A weak answer jumps to Elasticsearch without questioning whether it's necessary.

Live troubleshooting:

"Your RDS instance reached 80% CPU and your auto-scaling ECS tasks can't connect to the database for 2 minutes every hour, always at the same time. Walk me through how you'd debug this."

The candidate should think:

• Correlate ECS connection spikes with RDS load
• Check for slow queries via Enhanced Monitoring
• Look for connection leaks (app not closing connections)
• Consider if something else is running at that hour (backups, reports)

Azure assessment framework

Skills to test

1. Identity and access (Azure AD / Entra)

   • Managed identity vs. service principals
   • Role-based access control (RBAC) with custom roles
   • Conditional Access policies for security

2. Networking and security

   • NSGs, UDRs, and peering
   • Azure Firewall vs. NVA
   • Private endpoints and service endpoints

3. Storage and compute

   • Managed disks, premium SSD, ephemeral OS disks
   • App Service vs. Container Instances vs. AKS
   • Azure SQL failover groups vs. read replicas

4. Observability

   • Application Insights vs. Azure Monitor
   • Log Analytics queries (KQL)
   • Diagnostic settings and retention

Assessment question examples

Take-home scenario:

"You're migrating an on-premises application to Azure. The app requires a fixed IP, has strict compliance requirements, and must be isolated from other tenants. Design the network and identity architecture."

Good answers include:

• Dedicated subnet with NSG rules
• Private endpoint for databases
• Managed identity for app authentication
• Network policies for traffic control

Live conversation:

"Your Azure Web App is slow but Azure Monitor shows normal CPU. Latency spiked after deploying new code. What do you check first?"

The candidate should think about:

• Application Insights traces and dependencies
• App Service plan CPU vs. app-level bottlenecks
• Code-level changes (new queries, blocking calls)
• Database connection pooling
• Third-party dependency latency

GCP assessment framework

Skills to test

1. IAM and organization hierarchy

   • Service accounts and key management
   • Custom roles and least privilege
   • Organization policies and constraints

2. Compute and orchestration

   • Compute Engine zones and regions
   • GKE cluster design and autoscaling
   • Cloud Run for serverless workloads

3. Data and storage

   • Cloud Storage buckets, versioning, lifecycle
   • BigQuery for data warehousing
   • Cloud SQL and Cloud Spanner failover

4. Observability (Google Cloud Operations)

   • Cloud Logging and log routing
   • Cloud Trace and profiling
   • Cloud Monitoring and custom metrics

Assessment question examples

Take-home scenario:

"You have a batch processing job that runs daily on Compute Engine, processes 1TB of data, and takes 2 hours. You want to reduce cost while maintaining the SLA. Design a solution."

Good answers might consider:

• Preemptible VMs (cost-effective, acceptable for batch)
• BigQuery for querying (faster than Compute Engine processing)
• Cloud Run if parallelizable
• Spot VMs for fault-tolerant workloads

Live debugging:

"Your Cloud Run service is timing out on some requests. Logs are clean. P99 latency went from 50ms to 2s after a deploy. Trace this."

The candidate should think:

• Check traces for which operation is slow
• Investigate new dependencies (database, external APIs)
• Check database connection limits
• Review memory allocation (does Cloud Run need more?)
• Look for cold start patterns

Cross-platform assessment (platform-agnostic)

For candidates who've used multiple clouds, test transferable skills:

Question:

"You've architected systems on AWS and Azure. What are the design patterns that transfer between them? What's fundamentally different?"

Good answer acknowledges:

• Core concepts (VPCs, IAM, managed databases) are similar
• Implementation details differ significantly
• Cost models differ (reserved instances vs. commitment)
• The learning curve for each platform is shallow once you understand the first

This signals they're not dogmatic and can move between clouds.

Assessment structure for cloud hiring

1. 30-minute take-home (platform-specific scenario with constraints)
2. 30-minute live troubleshooting (use their platform of expertise)
3. 15-minute architecture conversation (test whether they understand platform trade-offs)

When to hire multi-cloud vs. platform-specialist

Hire multi-cloud generalists if:

• You're migrating between platforms
• You need flexibility to negotiate vendor terms
• You're building portable infrastructure-as-code

Hire platform-specialists if:

• Your platform decision is locked in
• You need deep operational knowledge of that platform's edge cases
• You're optimizing costs on a specific cloud

Most healthy engineering orgs need both: a few deep specialists and more generalists who can operate across clouds.

Building fair assessments across platforms

If you're assessing both AWS and Azure candidates, structure your questions consistently. Ask each candidate to design the same system, just on their platform. Then compare: do the architectures have equivalent trade-off thinking? Or did one candidate take shortcuts?

Ready to build your cloud hiring assessment? Explore DevOps and Cloud Engineering resources and how to interpret assessment results.