DSA interview questions

You have solved a few hundred problems and still feel unready, which usually means you have been practicing the wrong half of the exercise. This section names what is being scored, so the practice can match it.

A live data structures and algorithms round measures four things at once: whether you recognize which pattern the problem belongs to, whether you can state the approach before writing code, whether the implementation is correct at the boundaries, and whether you can reason about complexity out loud. Candidates optimize the third and neglect the second, then get marked down for a correct solution that arrived without explanation.

The recognition step is why pattern practice beats volume. Problems are rarely novel; they are recombinations. Once you can say this is a sliding window over a frequency map within the first minute, the rest is execution. Related question banks by language and role sit on the interview questions hub.

These patterns account for the majority of what gets asked. Learn to recognize the signal that points at each one.

• Two pointers. Sorted input, pair sums, in-place partitioning, palindrome checks.
• Sliding window. Longest or shortest contiguous subarray under a constraint.
• Hash map counting. Anagrams, duplicates, frequency comparisons, first unique element.
• Binary search. Sorted arrays, and search on the answer space when the array is not sorted.
• Breadth-first and depth-first search. Trees, grids, connected components, shortest unweighted path.
• Heap and priority queue. Top-k problems, merging sorted streams, running medians.
• Intervals. Merging, inserting, and detecting overlap after sorting by start.
• Dynamic programming. Overlapping subproblems: climbing stairs, coin change, edit distance, subsequences.
• Backtracking. Permutations, combinations, subsets, constraint puzzles.
• Graph algorithms. Topological sort, union find, weighted shortest path.

Representative prompts, phrased the way interviewers phrase them:

• Find two numbers in an array that sum to a target, then do it without extra space.
• Return the length of the longest substring without repeating characters.
• Given a rotated sorted array, find a target in logarithmic time.
• Merge overlapping intervals and explain why sorting is worth its cost.
• Invert a binary tree, then find its maximum depth.
• Validate a binary search tree, and say what a naive check misses.
• Count islands in a grid, then say how you would handle a grid too large for memory.
• Find the k most frequent elements and justify your data structure.
• Compute the minimum number of coins for an amount, and state the recurrence.
• Detect a cycle in a linked list, then return the node where it begins.
• Serialize and deserialize a binary tree.
• Given course prerequisites, decide whether the schedule is possible.

A new graduate interviewing for a backend role is given the substring problem and starts typing immediately. The code is nearly right, but the interviewer spends the round asking what it does, and the score reflects the silence rather than the bug. A candidate who spends forty seconds saying window over a character map, expand right, contract left on a duplicate, track the maximum has already passed the part that is hardest to recover.

Solving silently builds the wrong reflex. The live round requires narration and code at the same time, and that is a separate skill from either one alone.

Change the drill rather than the problem set. Before writing anything, say the pattern, the approach, and the expected complexity out loud. Write the code while continuing to narrate. When you finish, state the complexity again and name one edge case you handled and one you would ask about. Doing this on ten problems builds more interview capability than solving fifty in silence.

A mock interview session supplies the follow-up questions, which is the part that cannot be rehearsed alone, and gives you a recording of your own narration to review. One honest limit applies throughout: a typed, in-browser coding assessment with proctoring is not a conversation, and no live assistant belongs in one. The interview types hub covers which coding formats are live and which are automated.