Checkpoint: midpoint mock round
How to run these rounds
Read this before starting. The rounds are timed, and the timing is the point.
Why this one is different
The last two checkpoints tested whether you could recognize a pattern without being told. This one tests whether you can do it on a clock, which is a separate skill and the one the actual interview measures.
Nothing here is harder than what you have already done. The difference is that you should time yourself and stop when the time is up, whether or not you are finished.
Key idea
Two rounds
Each round is one warm-up problem and one main problem, on a 45 minute clock for the pair. That is the shape of a typical screen: something short to settle in, then something with substance.
Set a timer. When it goes off, stop, even mid-function. What you had at 45 minutes is the honest signal; what you have at 90 is not.
Why it works
Talk through it, even alone
Say what you are doing as you do it. Restate the problem. Name the brute force and its cost. Say which pattern you think applies and why. Say what you are about to test.
This feels ridiculous with nobody in the room and it is the single highest-value thing to practice, because an interviewer is scoring your reasoning at least as much as your code. A silent correct solution scores worse than a narrated one, and doing it under time pressure for the first time in the real interview is the wrong plan.
Tip
If you stall
Give yourself five minutes of genuine stuckness, then take a nudge rather than burning the clock. Running out of time with nothing written is a worse outcome than finishing with one hint.
Note when you took it and what unstuck you. That note is the most useful thing this checkpoint produces.
Key idea
What is in scope
Everything through Unit 9: hashing, two pointers, windows, stacks and monotonic stacks, binary search, intervals, linked lists, recursion and trees, and heaps.
Nothing below requires backtracking, graphs, or dynamic programming. If you find yourself reaching for one of those, you have almost certainly misread the problem.