This page needs a newer browser to run its tools. You can still read everything here.

How to approach a coding round interview questions

Interview questions on how to approach a coding round: short warm-ups, then what each harder question tests, the first step of a good answer and its follow-ups.

21 questions: 7 warm-ups and 14 with model answers · Levels: Fresher to senior

Warm-up questions

Short questions with short answers, each answered in full.

Fresher

Besides working code, what does an interviewer score in a live coding round?

Answered in the lesson Coding rounds: formats and what gets scored
Fresher

Why state a brute force before you optimise in a coding interview?

Answered in the lesson A six-step method for any coding problem
Fresher

How do you turn a problem's input limits into a target time complexity?

Answered in the lesson Read constraints to pick the target complexity
Fresher

How do you get from a brute force to a faster solution without guessing?

Answered in the lesson From brute force to optimal: the usual moves
Fresher

Which edge cases do you check before you start coding?

Answered in the lesson Edge cases and tests before you code
Fresher

What is a loop invariant, and how do you use one in a coding round?

Answered in the lesson Dry runs and loop invariants
Fresher

What do you do in a live coding round when no approach comes to mind?

Answered in the lesson Getting unstuck: heuristics and using hints

Questions with model answers

What each question tests, the first step of a good answer and the follow-ups to expect.

Fresher to mid

Your practised Python solution imports itertools.batched and uses heapq.heappush_max for a max-heap. The assessment platform's page lists CPython 3.10. What happens when you submit, how could you have caught it before the test, and what do you write instead?
What it tests

Knowing that a missing function fails before any test runs, checking a platform's versions with a probe, and rewriting with older parts of the standard library.

First step

Say what fails and when. batched is new in Python 3.12 and heappush_max in 3.14, so on 3.10 the import stops the program at start-up and every hidden test scores nothing, however good the algorithm is. Then explain how a probe run in a practice test would have shown it.
Follow-ups
- How do you cut a list into batches of k items on Python 3.10?
- How do you get a max-heap from heapq without the new functions?
- What would you check about JavaScript on the same platform?
From the lesson Coding rounds: formats and what gets scored
Model answer, Pro. Model answers open with any pass; prices are in the notice on this page.
Opening the model answers with your pass…
Model answers open with any MySmartCoPilot pass.
Pro pass · 1 week₹79$2.99
30 days₹179$5.99
1 year₹1,699$45
One payment; nothing renews unless you choose to. The price shown is the total you pay: no fees or charges are added at checkout.
Passes can’t be bought right now, so the model answers can’t be opened at the moment. If you have a pass, enter your licence key.
Choose a pass I have a licence key
It also opens every other in-depth part in Learn and every Pro tool. Sales are final, except where the law gives you a right to withdraw, for example in the EU and the UK (refund policy).
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

Next week you have three coding rounds: an online assessment on a platform you have never used, a live round in a shared editor, and a round that allows an AI assistant. How do you prepare for each, and what do you do differently in the first ten minutes of each?
What it tests

Knowing what each format scores, turning that into preparation, and changing the opening habits: checking the platform, saying the plan aloud, and splitting the task before asking the assistant.

First step

Start from who judges each round: hidden tests, an interviewer, or an interviewer who also reads the assistant's code. Then name the habit each judge rewards most: a correct submission early, evidence said aloud, and generated code that you have read and tested.
Follow-ups
- What do you practise on the unfamiliar platform before the assessment?
- What do you say in the first two minutes of the live round?
- How do you check the assistant's code without stalling the round?
From the lesson Coding rounds: formats and what gets scored
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

Two minutes into a live round, the interviewer says: "Skip the brute force, I want the optimal solution." How do you respond, and what do you still say or check before you write the optimal code?
What it tests

Adapting the method without losing what it protects: confirming the problem in one sentence, stating costs and keeping a way to check the code, while respecting the request.

First step

Agree, then shrink step 3 to one sentence that names the simplest correct method and its cost, so a misread can still be caught: "Every pair works and is O(n²); I'll go straight to sorting." Then say the one claim the fast plan rests on before you type.
Follow-ups
- What if your one-sentence brute force shows that you misread the problem?
- How do you test the optimal code without a brute force to compare against?
- When is it right to skip the brute force altogether?
From the lesson A six-step method for any coding problem
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

Your clash check sorts slots by start and compares each slot with the one before it. The interviewer asks you to prove that no clash between slots far apart in the sorted order can be missed, then changes the rule so that touching slots clash too. What changes in the proof and in the code?
What it tests

Proving the neighbour property of sorted intervals, adapting it when the clash rule changes from strict to non-strict, and matching the comparison in the code to the rule.

First step

Take any clashing pair, call A the one that starts first and B the slot right after A in sorted order. Show that B must clash with A using only the start times and A's end, then look for the one inequality that depends on whether touching counts.
Follow-ups
- Does the proof need the sort to break ties by end time?
- How do you report every clashing pair instead of a yes or no, and what does that cost?
- What changes if slots are written as whole minutes with the end included, such as 540 to 599?
From the lesson A six-step method for any coding problem
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

In a live round the interviewer states a problem but gives no input limits. Which limits do you ask for, how do you turn the answers into a target complexity, and what do you do if the interviewer says "as large as you like"?
What it tests

Asking for the limits that decide the algorithm, converting them into a step budget aloud, and choosing a sensible target when the limits stay open.

First step

Ask for the largest input size, the range of the values and the number of queries or test cases, then turn the size into a step count aloud: "If n reaches 2 × 10⁵, every pair is about 2 × 10¹⁰ steps, so I am aiming for O(n log n)."
Follow-ups
- What do you do with an answer such as 5 × 10⁶ that you cannot place at once?
- How does the range of the values change the plan, apart from the size?
- Is a brute force acceptable when the interviewer says n is small?
From the lesson Read constraints to pick the target complexity
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

Your O(n log n) Python solution takes 1.6 seconds on the largest input you built, n = 10⁶, and the limit is 1 second. How do you decide whether the complexity class or the constant factor is the problem, and what do you change before you switch languages?
What it tests

Measuring growth by doubling n, estimating the step count against CPython's rate, and removing constant costs such as per-item Python work, slow input reading and printing.

First step

Time the solution at n = 500,000 and at n = 1,000,000. A little over twice the time means the class is right and the constant is the problem; about four times means hidden quadratic work, which no tuning will save.
Follow-ups
- Where does the time of a Python solution usually go, apart from the algorithm?
- What does n log₂ n come to at n = 10⁶, and what does that say about the budget?
- When is switching to another language the right call?
From the lesson Read constraints to pick the target complexity
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

For n ≤ 10⁵ daily prices, count the pairs of days whose prices differ by at most d. Your brute force compares every pair. Which move removes its repeated work, what does the result cost, and how do you show that it counts every pair exactly once?
What it tests

Naming the waste of a pair loop, choosing a sort with two pointers, giving the cost, and arguing that the set of pairs does not change when the prices are sorted.

First step

Say the waste: for each day, the inner loop scans every other day for prices within d. After sorting, the prices within d above a value sit in one unbroken stretch to its right, so a second pointer can find where that stretch ends without starting again.
Follow-ups
- Why is sorting allowed here, when it broke the best-profit problem?
- How large can the answer get, and which integer type must hold it?
- What changes if the two days must also be at most w days apart?
From the lesson From brute force to optimal: the usual moves
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

A service answers 10⁵ questions of the form "what was the best profit from one buy and a later sale between day l and day r?" over 10⁵ days of prices. One pass per question is too slow. What do you precompute, how do you combine it, and what does each question cost?
What it tests

Extending a running-value move to ranges with a summary that can be merged, writing the merge in the right order, and comparing build and query costs with a scan per question.

First step

Price the simple plan first: a pass per question is up to 10⁵ steps, so 10⁵ questions cost about 10¹⁰. Then ask what a range must know about its left and right halves to give its best profit without looking inside them again.
Follow-ups
- Why does a sparse table of minimums and maximums not answer it on its own?
- What does the merge need when the buy is in the left half and the sale in the right?
- What changes if a day's price can be corrected between the questions?
From the lesson From brute force to optimal: the usual moves
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

A problem asks for the largest sum of two values from a list of up to 2 × 10⁵ integers, each between −2 × 10⁹ and 2 × 10⁹. Your solution passes the samples but fails a hidden test. Which edge cases do you suspect, and which tests do you write before you submit again?
What it tests

Reading the limits for overflow, negative values and the smallest sizes, and turning each suspicion into a test with its expected answer before changing the code.

First step

Read the limits as edge cases. Each value fits in a 32-bit int, but two values of 2 × 10⁹ add up to 4 × 10⁹, past its 2,147,483,647, and a list of negative values breaks code that starts its best sum at 0. Write those tests with their answers first.
Follow-ups
- What are the largest and the smallest possible answers, and which type holds them?
- What must the function return for a list of exactly two values?
- How would a stress test have found the bug if the limits had not pointed at it?
From the lesson Edge cases and tests before you code
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

You change one comparison in your solution from < to <=, and all 20 of your tests still pass. The interviewer asks what that says about your tests and what you would do next.
What it tests

Telling an equivalent change from a missing test, building the input on which two versions differ, and judging a test suite by the mistakes it catches.

First step

Say what the change can affect: < and <= differ only when the two sides are equal. Build an input with a tie at that comparison. If the two versions answer it differently, a test is missing; if no input can make them differ, the change is equivalent.
Follow-ups
- How do you decide quickly whether the two versions can ever differ?
- Which tests do you add for a comparison inside a loop that keeps the best value so far?
- How many such changes would you try before you trust a test suite?
From the lesson Edge cases and tests before you code
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

Two pointers start at both ends of a sorted list and move inwards to find two values that add up to a target. State the invariant that proves the loop cannot skip the pair, and use it to choose the loop condition and both updates.
What it tests

Writing an invariant for a two-pointer scan, arguing maintenance for each move, and deriving the loop condition and termination from it instead of from memory.

First step

Say what the pointers promise about the pairs already ruled out: no pair that uses a value left of lo, or a value right of hi, adds up to the target. Then check that moving lo after a sum that is too small keeps the promise, because a[lo] plus any value up to a[hi] is no larger.
Follow-ups
- Why is it safe to move lo when the sum is too small?
- Why is the loop condition lo < hi and not lo <= hi?
- How many passes can the loop make, and why must it stop?
From the lesson Dry runs and loop invariants
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

A teammate's search for the first position whose value is at least x in a sorted list passes every sample test but runs out of time on one hidden test. You may not run it. How do you find the bug with an invariant and a dry run, and which inputs do you trace first?
What it tests

Reading a time-out as a loop that stops making progress, writing the invariant the code relies on, and choosing the smallest inputs where a boundary update changes nothing.

First step

A time-out without a wrong answer points at a pass that changes neither lo nor hi. Write the invariant the code seems to keep, then trace ranges of two positions, where mid equals lo, and check that every branch moves lo or hi there.
Follow-ups
- Which two-position input shows the hang when the code writes lo = mid?
- How do you prove that the loop ends without running it?
- What tests would you add so that the bug cannot come back?
From the lesson Dry runs and loop invariants
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Fresher to mid

For 10⁵ questions of the form "total sales from day l to day r" over 10⁵ days, your brute force adds up the days of each range. The interviewer asks: "What does your brute force compute twice?" How do you turn that hint into a plan, and how do you code it without losing your working solution?
What it tests

Restating a hint as an operation on the data, choosing the value it says to store, and changing one part of the program while the brute force stays as a checker.

First step

Restate the hint before coding: overlapping ranges add up the same days again and again, so store the running totals of every prefix once and answer each range as the difference of two of them. Check that operation by hand on a five-day example first.
Follow-ups
- What does the plan cost, compared with the brute force?
- Where is the off-by-one trap in the array of prefix totals, and how do you avoid it?
- How do you check the new version before it replaces the old one?
From the lesson Getting unstuck: heuristics and using hints
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes
Mid to senior

You have a fast solution and a slow, plainly correct one for a problem with up to 10⁵ values between 1 and 10⁹. Design the stress test you run before you submit: the inputs, their sizes, the seed, what it prints on a mismatch, and when you stop.
What it tests

Designing a stress test that is quick, repeatable and likely to reach the edge cases, and turning its first failure into a small test that stays.

First step

Keep the inputs small so that the brute force stays quick, try the shortest first, and draw values from a narrow range so that ties and repeats happen, with a few lists at the extremes. Fix the seed, and on a mismatch print the input itself.
Follow-ups
- Why small values as well as small sizes?
- What should the test print, and why the input rather than only the seed?
- How many cases are enough, and how do you know the test compared real answers?
From the lesson Getting unstuck: heuristics and using hints
Model answer, Pro.
In the answer:
- Model answer
- Follow-up answers
- Common mistakes
- Rubric
- Variants
- about 3 minutes

The lessons of this module

Coding rounds: formats and what gets scored Free lesson, Pro answers: 20 minutes
A six-step method for any coding problem Free lesson, Pro answers: 20 minutes
Read constraints to pick the target complexity Free lesson, Pro answers: 15 minutes
From brute force to optimal: the usual moves Free lesson, Pro answers: 20 minutes
Edge cases and tests before you code Free lesson, Pro answers: 20 minutes
Dry runs and loop invariants Free lesson, Pro answers: 20 minutes
Getting unstuck: heuristics and using hints Free lesson, Pro answers: 20 minutes

All Coding Interview Patterns interview questions · The Coding Interview Patterns track

Warm-up questions

Besides working code, what does an interviewer score in a live coding round?

Why state a brute force before you optimise in a coding interview?

How do you turn a problem's input limits into a target time complexity?

How do you get from a brute force to a faster solution without guessing?

Which edge cases do you check before you start coding?

What is a loop invariant, and how do you use one in a coding round?

What do you do in a live coding round when no approach comes to mind?

Questions with model answers

Your practised Python solution imports itertools.batched and uses heapq.heappush_max for a max-heap. The assessment platform's page lists CPython 3.10. What happens when you submit, how could you have caught it before the test, and what do you write instead?

Next week you have three coding rounds: an online assessment on a platform you have never used, a live round in a shared editor, and a round that allows an AI assistant. How do you prepare for each, and what do you do differently in the first ten minutes of each?

Two minutes into a live round, the interviewer says: "Skip the brute force, I want the optimal solution." How do you respond, and what do you still say or check before you write the optimal code?

Your clash check sorts slots by start and compares each slot with the one before it. The interviewer asks you to prove that no clash between slots far apart in the sorted order can be missed, then changes the rule so that touching slots clash too. What changes in the proof and in the code?

In a live round the interviewer states a problem but gives no input limits. Which limits do you ask for, how do you turn the answers into a target complexity, and what do you do if the interviewer says "as large as you like"?

Your O(n log n) Python solution takes 1.6 seconds on the largest input you built, n = 10⁶, and the limit is 1 second. How do you decide whether the complexity class or the constant factor is the problem, and what do you change before you switch languages?

For n ≤ 10⁵ daily prices, count the pairs of days whose prices differ by at most d. Your brute force compares every pair. Which move removes its repeated work, what does the result cost, and how do you show that it counts every pair exactly once?

A service answers 10⁵ questions of the form "what was the best profit from one buy and a later sale between day l and day r?" over 10⁵ days of prices. One pass per question is too slow. What do you precompute, how do you combine it, and what does each question cost?

A problem asks for the largest sum of two values from a list of up to 2 × 10⁵ integers, each between −2 × 10⁹ and 2 × 10⁹. Your solution passes the samples but fails a hidden test. Which edge cases do you suspect, and which tests do you write before you submit again?

You change one comparison in your solution from < to <=, and all 20 of your tests still pass. The interviewer asks what that says about your tests and what you would do next.

Two pointers start at both ends of a sorted list and move inwards to find two values that add up to a target. State the invariant that proves the loop cannot skip the pair, and use it to choose the loop condition and both updates.

A teammate's search for the first position whose value is at least x in a sorted list passes every sample test but runs out of time on one hidden test. You may not run it. How do you find the bug with an invariant and a dry run, and which inputs do you trace first?

For 10⁵ questions of the form "total sales from day l to day r" over 10⁵ days, your brute force adds up the days of each range. The interviewer asks: "What does your brute force compute twice?" How do you turn that hint into a plan, and how do you code it without losing your working solution?

You have a fast solution and a slow, plainly correct one for a problem with up to 10⁵ values between 1 and 10⁹. Design the stress test you run before you submit: the inputs, their sizes, the seed, what it prints on a mismatch, and when you stop.