// Cloud / Mental Model / v1

AWS, mapped.

11 functional groups. ~70 services. Built for fast recall in interviews and real architecture decisions, not for marketing taxonomy.

11  categories
70+  services
★  must-know
ComputeStorageDatabaseNetworkingSecurity & IAMDevOps & IaCObservabilityIntegrationAnalyticsAI / MLGovernance

Compute

01 / RUN CODE

Storage

02 / PUT BYTES

Database

03 / PERSIST STATE

Networking

04 / MOVE PACKETS

Security & IAM

05 / GUARD GATES

DevOps & IaC

06 / SHIP CODE

Observability

07 / SEE TRUTH

Integration

08 / WIRE SYSTEMS

Analytics

09 / QUERY DATA

AI / ML

10 / TRAIN MODELS

Governance

11 / OWN ACCOUNTS

// The 20% That Drives 80%

  1. Compute + Containers — EC2, Lambda, ECS vs EKS, Fargate tradeoffs
  2. Networking — VPC layout, subnets, ALB vs NLB, Route 53 policies
  3. IAM — roles vs users, assume-role, instance profiles, least privilege
  4. Storage + DB — S3 classes, RDS vs Aurora vs DynamoDB
  5. DevOps + Observability — Terraform, CloudWatch, X-Ray, CodePipeline

// Pre-Mortem — Why This Fails

Memorizing names without tradeoffs— interviewers ask “S3 vs EFS for X,” not “what is S3.” Pair every service with its alternative.

Skipping IAM— #1 source of real-world failures, #1 deep-dive area for platform EM interviews.

Over-rotating on AI/ML— for frontend platform EM, ML is decorative. Cut it to 10% of study time.