Interview prep · field manual

Stop memorizing.
Start recognizing.

Every Grind 75 problem is a costume worn by one of a small number of patterns. Learn to spot the pattern from the problem's signals, reach for the matching template, and adapt. This is the recognition map — 20 patterns, each with the tells that trigger it, an illustrated idea, a Python template, and a flagship Grind 75 solution.

Core patterns

75+

Mapped problems

∞

Variations they spawn

00 · Orientation

How to actually use this

The list's own author says it plainly: the goal is to get good at question patterns, not to memorize answers. Treat each card as a lens, not a flashcard.

Read the "use when" cues first. In an interview you get a problem, not a label. Train the reflex: "sorted array + find a pair → two pointers". The cues are the most valuable part of every card.

Type the template from memory, then adapt. Don't copy. The muscle memory of writing the skeleton (the two-pointer loop, the BFS queue) is what survives interview stress.

Solve the flagship, then the "related" problems. Same pattern, new disguise. Doing 3–4 variants is how a pattern becomes automatic. Always say your complexity out loud.

The trap: grinding 300 problems by rote. Beyond the first ~1000 LeetCode IDs, almost everything is a rehash of these same patterns. Depth on the pattern beats breadth on the problems.

01 · Triage

Pattern decision guide

When you read a fresh problem, scan for these signals. They're not laws — many problems mix patterns — but they're the fastest first guess.

If the problem says / has…

First reach for

No pattern matches that filter. Try another keyword.

22 · Reference

Big-O cheatsheet

What interviewers expect you to recognize on sight. "Acceptable" assumes typical constraints (n up to ~10⁵–10⁶).

Complexity	Name	Typical source	Verdict at n≈10⁵

★

Constraint-reading trick. The input size tells you the target complexity. n ≤ 20 → exponential/backtracking is fine. n ≤ 2000 → O(n²) ok. n ≤ 10⁵ → you need O(n log n) or better. n ≤ 10⁹ → O(log n) — think binary search.

23 · Reference

Data structure operations

The Python tools behind every pattern, and what each operation costs.

Structure	Python	Access / Search	Insert	Delete	Notes

Now go disguise-hunting.

This compendium is your recognition map. The real skill is built one problem at a time on the official list — techinterviewhandbook.org/grind75 — and pattern banks like SeanPrashad/leetcode-patterns.

Pick a pattern. Do its flagship. Do its variants. Repeat until the disguise is transparent.