Loading…
Cloud engineer, infrastructure engineer, platform engineer and DevOps engineer interviews at Google, Amazon, Microsoft and the companies that run FAANG-style loops assess whether you can build and run the layer other engineers work on top of: account structure and network topology, identity, infrastructure as code with its state and drift, the delivery pipeline, progressive rollout and rollback, secrets and configuration, the observability that tells you it is working and the backups that get it back when it is not. The design round is an operational one — a pipeline, a rollout, a secrets workflow — far more often than a product system, and interviewers probe the operational difficulty (state locking, blast radius, the rollback path) rather than the diagram. This plan sequences the site's operational design questions, cloud architecture questions, discussion-round answers, concept pages and quizzes into six weeks, easiest first, with a free week one so you can judge the format before paying for anything.
Built for someone with a few weeks before a cloud, infrastructure, platform or DevOps engineer loop who has run infrastructure as code and a delivery pipeline in production. If your loop is an SRE loop, the SRE plan adds SLOs and capacity work on top of this one; if it is a cloud architect loop, the cloud architect plan swaps the operational round for service selection and the discussion week.
Twelve weeks, 43 steps, about 20 hours. Week one is free with no account.
Learn what this loop actually weighs — the operational difficulty, not the diagram — and finish two real designs: how a platform knows a thing is healthy, and which compute model to put it on. Everything this week is free.
Six weeks at roughly three to four hours a week, or about 20 hours in total. The plan compresses to as few as two weeks by dropping the optional steps first, and stretches to twelve; the core steps stay in every version.
Week one is entirely free with no account: the infrastructure-as-code and deployment-strategies discussion questions, the health checking design, the compute-model decision and the quiz. Later weeks use Pro questions for the operational design round; every discussion-round answer, blog post, concept page, quiz and tool stays free throughout.
Progress is derived from questions you have completed and quizzes you have taken, not from a stored position in the plan. If you finished the CI/CD pipeline design last month, the plan already shows it done. Reading steps are not tracked, so the percentage counts questions and quizzes only.
Week one is free. From week 2 the plan uses 12 Pro questions and 3 Pro behavioral guides; the posts, concept pages, quizzes and tools stay free.
See Pro pricingWeeks 1, 2 of the six-week plan, combined.
Design the boundaries other teams build inside — account structure as the blast radius, identity as the perimeter — and the pipeline that delivers into them.
The round this loop is decided on. Design the rollout that rolls itself back, the secret that rotates without an outage, the config change that cannot take down every region at once — and say the failure path out loud each time.
Rollout with automatic rollback, secrets rotation, configuration distribution: the design questions this loop asks almost verbatim, where the whole interview is the failure path — what happens when the rollout is bad, when a secret leaks, when a config change is wrong. Easiest first; a new operational question lands here on its own.
Weeks 3, 4 of the six-week plan, combined.
Rollout with automatic rollback, secrets rotation, configuration distribution: the design questions this loop asks almost verbatim, where the whole interview is the failure path — what happens when the rollout is bad, when a secret leaks, when a config change is wrong. Easiest first; a new operational question lands here on its own.
Know how the platform tells you it is working, and how you find out when it is not: the monitoring design, the tracing design, and the structured investigation of an intermittent failure.
Weeks 4, 5 of the six-week plan, combined.
The half most preparation skips: the platform's stateful services — restoring against a stated recovery objective, moving a live database, running a cache tier — and explaining the bill.
Weeks 5, 6 of the six-week plan, combined.
The coding round here is a working script under time, not an algorithm exam: parse, count, dedupe, look up. Arrays, hash maps, strings and binary search cover it; take the easiest first and stop when they feel routine.
The lighter coding round, the behavioral round where platform engineers are asked about the incident and the migration they ran, then one full design under time.
The coding round here is a working script under time, not an algorithm exam: parse, count, dedupe, look up. Arrays, hash maps, strings and binary search cover it; take the easiest first and stop when they feel routine.
The lighter coding round, the behavioral round where platform engineers are asked about the incident and the migration they ran, then one full design under time.
The coding round here is a working script under time, not an algorithm exam: parse, count, dedupe, look up. Arrays, hash maps, strings and binary search cover it; take the easiest first and stop when they feel routine.
The lighter coding round, the behavioral round where platform engineers are asked about the incident and the migration they ran, then one full design under time.
Typically an operational design round — a delivery pipeline, a rollout, a secrets or configuration workflow, sometimes a landing zone — a discussion round on infrastructure as code, deployment strategies, security and observability, a troubleshooting question, a lighter coding round, and a behavioral round. Platform engineer and DevOps engineer loops are the same shape under a different title.
They overlap on observability and the operational design round, because the loops do. The SRE plan adds service level objectives, error budgets and capacity estimation with the numbers-first method; this plan spends that time on the platform itself — account structure, infrastructure as code, the pipeline, secrets and configuration, recovery. If your job ad says SRE, take that one; if it says cloud, infrastructure, platform or DevOps, take this.
No, deliberately. Every question names a self-managed and a managed option for each capability and treats the providers as equivalents, because a candidate is rarely interviewing at the cloud they last used and interviewers assess the reasoning, not the product names.