Practice for interviews.
78 problems across the six rounds a modern loop is assembled from. Each one gives you the question, a clock, three hints you have to ask for, a worked solution written by a person, and a rubric you score yourself against. Free, no account, nothing graded.
Forty-five minutes, an ambiguous product, and a whiteboard. The round that decides your level.
- Design a URL shortener
- Design a rate limiter
- Design a news feed
- Design a chat system
- Design a web crawler
- Design a notification system
- + 10 more
Start with Design a URL shortener. You are given a product in one sentence and asked to design it. Expect five minutes on requirements, ten on a first design, and then the half that actually counts: the interviewer picks the weakest part and pushes.
One function or one component, someone watching, and the clock on.
- Implement debounce, then make it cancellable
- Build an event emitter
- Deep clone a value
- Run tasks with a concurrency limit
- Retry with exponential backoff and jitter
- LRU cache in O(1)
- + 12 more
Start with Implement debounce, then make it cancellable. Thirty to forty-five minutes in a shared editor. You will usually be asked to state the approach before you type, and almost always to handle the case you skipped.
Classes, interfaces and state machines. The round where "it depends" has to become a diagram.
- Design a parking lot
- Design a vending machine
- Design an elevator system
- Design an expense-splitting app
- Design an in-memory file system
- Design a logging library
- + 6 more
Start with Design a parking lot. Forty-five minutes. You are asked to model a small system in objects, entities, interfaces, the state machine, the concurrency, and usually to write the core class or two in real code.
Two minutes, specific, with a real cost in it. The round people prepare least and lose most on.
- Tell me about yourself
- Tell me about a mistake that cost something
- Tell me about a time you disagreed with a decision
- What is the hardest technical problem you have solved?
- Tell me about a deadline you were going to miss
- Tell me about feedback you had to give someone
- + 6 more
Start with Tell me about yourself. Thirty to forty-five minutes of "tell me about a time". Expect four or five questions, each followed by two or three probes that are the actual interview.
You bring the model. You are scored on direction and verification, never on typing.
- Build a feature with a model, out loud
- Review generated code that is subtly wrong
- Catch an API that does not exist
- Have a model write the tests, then find what it missed
- Find the security bug in generated code
- Refactor code you have never seen, with a model
- + 4 more
Start with Build a feature with a model, out loud. Sixty minutes, your own tools open, thinking out loud. The interviewer rarely reads your code closely; they are listening to what you asked for, what you expected back, and how you found out it was wrong.
A product goal, a label you have to invent, and everything that happens after the model is good.
- Design feed ranking
- Design a fraud detection system
- Design search ranking
- Design a recommendation homepage
- Design content moderation
- Design click-through rate prediction
- + 4 more
Start with Design feed ranking. Forty-five to sixty minutes. Expect the first ten on framing, what exactly is being predicted, from what, and what metric decides, and the rest on data, serving, and what goes wrong in production.
Free Everything in this room is free and works signed out. We run no AI here: every solution is pre-authored by a human, and the score at the end is your own.