Patterns Cheatsheet · Resources

1How to use this sheet

0/3▶

A cheatsheet is a recognition tool, not a solutions list. For each pattern, learn the signal that triggers it and the template — then the specific problem barely matters.

The drill

Read a problem and, before coding, say out loud which pattern it is and why.

Tech Interview Handbook — Coding cheatsheet — the "narrate before you code" checklist.

Memorise one clean template per pattern; adapt it, don't re-derive it under pressure.

NeetCode — Roadmap — patterns in dependency order, each with a template video.

For the deep per-pattern problem catalog, use the LeetCode — Patterns page.

Resources — LeetCode Patterns — 30 patterns, 90 Easy/Medium/Hard problems in our style.

2Two Pointers

0/3▶

Signal: a sorted array, or a question about pairs / palindromes / both ends. Turns many O(n²) brute forces into O(n), O(1) space.

Recognise & template

When: sorted input, or "find a pair/triplet", or compare from both ends.

Two Sum II — the sorted-array template.

Template: l=0; r=n-1; while l<r: move the pointer that improves the condition.

Valid Palindrome — both-ends variant.

Fast/slow variant: in-place partition — slow write-pointer, fast read-pointer.

Move Zeroes — the slow/fast template.

3Sliding Window

0/3▶

Signal: contiguous subarray/substring with a constraint ("longest/shortest/at most k"). Grow right, shrink left when the constraint breaks.

Recognise & template

When: "subarray/substring" + "longest/shortest/max/min" + a window constraint.

Longest Substring Without Repeating — the variable-window template.

Template: for r in range(n): add(r); while invalid: remove(l); l+=1; update answer.

Minimum Window Substring — shrink-to-minimise.

Fixed-size variant: add the new element, drop the one that fell out.

Maximum Average Subarray I — fixed window.

4Binary Search

0/3▶

Signal: sorted input, or you can phrase the answer as "smallest/largest x for which P(x) is true" (monotonic predicate). O(log n).

Recognise & template

Template (index): lo=0; hi=n-1; while lo<=hi: mid=(lo+hi)//2; ... — mind the boundaries.

Binary Search — write it once, cleanly.

Binary search on the answer: search a value range, checking feasibility at each mid.

Koko Eating Bananas — the "search the answer space" template.

Rotated array: at each mid, decide which half is sorted, then which side to keep.

Search in Rotated Sorted Array.

5BFS / DFS & Backtracking

0/3▶

Signal: trees, graphs, grids, or "generate all combinations/permutations". BFS for shortest unweighted path; DFS for reachability/paths; backtracking to enumerate with pruning.

Recognise & template

BFS template: queue + visited set; process level by level for shortest path.

Number of Islands — grid BFS/DFS.

DFS template: recurse on neighbours, mark visited; postorder to combine child results.

Max Depth of Binary Tree — postorder DFS.

Backtracking: choose → explore → un-choose; prune invalid branches early.

Subsets and Combination Sum — the decision-tree template.

6Heaps, Stacks & DP

0/4▶

The remaining high-value templates. Reach for a heap on "top-k / kth / merge k"; a monotonic stack on "next greater/smaller"; and DP when choices have overlapping subproblems.

Recognise & template

Top-K heap: keep a heap of size k (min-heap for k-largest). O(n log k).

Top K Frequent Elements.

Monotonic stack: push indices, pop while the new element breaks monotonicity.

Daily Temperatures — the next-greater template.

1D DP: define dp[i] and its recurrence; decide memoise vs tabulate; optimise space.

House Robber — the rob-or-skip recurrence.

2D DP: dp[i][j] over a grid or two sequences.

Longest Common Subsequence — the two-sequence template.