Arrays, Strings, And Matrix Fundamentals

What's being tested

These prompts test linear-time array processing, two-pointer in-place mutation, matrix indexing, BFS pathfinding, and streaming window aggregates. For an MLE, the interviewer is probing whether you can write production-safe primitives for feature streams, batched tensors, ranking lists, and grid/state traversal with clear O(n) / O(mn) complexity.

Patterns & templates

• Single-pass extrema tracking — for maxProfit(prices), maintain min_so_far and best; O(n) time, O(1) space.

• Reverse two-pointer merge — for merge(nums1, m, nums2, n), fill from the end to avoid overwriting; handle exhausted-array tails.

• Matrix diagonal traversal — compare matrix[i][i] or anti-diagonal matrix[i][cols-1-i]; validate rectangular shape and dimensions first.

• Bounds checking helper — centralize 0 <= r < rows and 0 <= c < cols; prevents repeated off-by-one bugs in grid logic.

• BFS for shortest path — use deque, visited, and parent map for path reconstruction; O(V+E) over grid cells.

• Sliding moving average — maintain deque plus running sum; update in O(1) per event, avoiding recomputation over the full window.

• Sliding window order statistic — for median/min/max, use heaps, monotonic queues, or balanced trees; average alone only needs sum.

Common pitfalls

Pitfall: Treating empty arrays, single-element arrays, or all-decreasing stock prices as afterthoughts; define return values before coding.

Pitfall: Merging sorted arrays from the front in-place, which overwrites unprocessed values in nums1.

Pitfall: Using DFS when the prompt asks for shortest path; BFS is the default for unweighted grid distances.

Practice these

The practice cards below cover the canonical variants — solve all of them and time yourself.