What every component does, and every feature explained.
Loading…
11 hand-crafted levels, from a basic webhook router to multi-region HA and RAG.
Code checks run before the AI, so an incomplete design is caught instantly — for free.
A persona-driven LLM scores your design on 5 axes and asks a probing follow-up.
Answer the interviewer or revise the canvas across up to 3 rounds per level.
Junior, Mid, and Senior grade against different bars — the same design scores differently.
Tune each node (replicas, size, multi-AZ…) — it changes cost, the stress test, and scoring.
A live monthly cost estimate makes over-engineering tangible.
Push a target RPS through your design to find the bottleneck node.
Relabel components as AWS / GCP / Azure products; the interviewer follows suit.
Assign regions to nodes for disaster-recovery and multi-region levels.
Practice an open-ended, timed scenario the AI generates on the spot.
Paste a real interview question or job posting — it becomes a playable scenario.
Track your stats and download a personalized, AI-written study report.
Play in English, Turkish, Spanish, French, or German — the interviewer speaks it too.
No account or login. Your name is paired with a random ID kept in this browser, so nobody can pull your history by guessing it. History & shared links live in Firestore; skip the name and nothing is saved.