Grade a whole design
One AI pass over your canvas + the rubric. Letter grade, per-criterion scores, flagged gaps, follow-up interviewer questions, study resources.
warming up your workspaceLoading
Interview prep for SRE, DevOps, Platform, and Cloud engineers. Real shells, real failure modes, real whiteboards that someone grades.
Real incidents don't have hint buttons. Neither do good interviewers.
Draw your architecture, get principal-level critique on capacity math, failure modes, and trade-off depth.
Pick a policy: cache-aside vs write-through. Also: TTL, eviction, and thundering-herd mitigation (singleflight / request coalescing).
The four-step loop you run on every problem, plus two AI-review modes that pressure-test your design after you draw it.
Pick a real infra problem — design, debug, or scenario
Draw your architecture, add constraints, annotate
AI reads your canvas + rubric, flags gaps
Per-track scores, streaks, next recommended drill
One AI pass over your canvas + the rubric. Letter grade, per-criterion scores, flagged gaps, follow-up interviewer questions, study resources.
Multi-turn review with a principal engineer. Three intensity levels — they press on the weakest parts until you sharpen or rebuild the argument.
106 problems across 6tracks, each with constraints → hints → solution. Progressive depth — we don't over-explain.
3 free problems per track. No card. No gamification.