Learn Algorithm Patterns, Not a Thousand Problems

Solving many problems can create familiarity without recall. In an interview, the useful skill is recognizing a constraint, selecting a pattern, and explaining why the complexity fits.

Organize by decision pattern#

Use a small set of families:

• hashing and frequency maps;
• two pointers and sliding windows;
• binary search on values or answers;
• monotonic stacks and queues;
• tree and graph traversal;
• heaps and top-K selection;
• intervals and sweep lines;
• dynamic programming by state and transition.

For each family, keep one canonical problem, one variation, and a short trigger list. For example, “contiguous range + constraint maintained while moving right” should prompt a sliding-window check.

Practice the explanation loop#

Before writing code, say:

1. the brute-force approach and its cost;
2. the repeated work or structure you can exploit;
3. the invariant maintained by the chosen pattern;
4. time and space complexity;
5. edge cases that could break the invariant.

That reasoning is often more valuable than arriving at a clever solution silently.

Revisit on a schedule#

After solving a problem, repeat it from a blank editor after one day, one week, and one month. If you cannot recover the pattern, write down the missed trigger—not the entire solution.

The target is not memorizing code. It is shortening the path from evidence to approach.