Coding assessment
The HackerRank Coding Assessment
HackerRank is the timed online coding screen that bank engineering, technology and quant teams use to filter applicants before any human looks at your CV. You solve algorithm problems in a browser code editor, answer computer-science multiple choice, and work through quantitative maths, all auto-graded against hidden test cases. Here is exactly how it works and how to pass it.
In short
A HackerRank coding assessment is a timed online test, usually 90 to 120 minutes, made up of two to three medium-to-hard algorithm and data-structure problems, a set of computer-science multiple-choice questions, and a quantitative maths section. Your code runs automatically against hidden test cases, so it must be correct and fast enough on large inputs, not just on the visible examples. Banks such as Goldman Sachs (Engineering), J.P. Morgan (Tech) and Morgan Stanley (Quant) use it as an early screen. You pass it by drilling data structures and algorithms, optimising for time complexity, and practising full timed sets in the real editor.
The basics
What it is
HackerRank is the dominant technical screening platform in technology recruiting, and banks use it the same way fast-growing tech firms do: as the first hard gate. Instead of a phone screen, you receive a link to a timed assessment, open a browser-based code editor, and work through a fixed set of tasks. There is no interviewer and no second chance once the timer starts, so almost everything rests on preparation and execution on the day.
For bank engineering, technology and Strats roles the assessment is deliberately broad. The headline component is two to three algorithm and data-structure problems that range from medium to genuinely hard. Alongside the coding you usually get a block of computer-science multiple-choice questions covering fundamentals, and a quantitative maths section that leans heavier for quant and Strats roles. The whole thing typically runs 90 to 120 minutes, and every section counts toward a single composite score.
The crucial difference from a take-home exercise is that your code is graded automatically against hidden test cases. You can see a handful of sample inputs, but the grader runs your solution against many more that you never see, including edge cases and large stress tests built to punish slow solutions. A brute-force answer that passes the visible samples can still fail most of the hidden cases because it times out, which is why algorithmic efficiency matters as much as raw correctness.
The configuration varies by firm and by year. Goldman Sachs runs its Engineering coding assessment on HackerRank and its setup is widely regarded as among the hardest on Wall Street, with demanding problems and tight time limits. J.P. Morgan uses HackerRank-style coding challenges across its technology hiring, and Morgan Stanley leans on it for quant and Strats roles where the maths and probability content carries more weight. The platform is identical in the UK and US, so the same preparation applies wherever you are applying.
What it measures
The dimensions under test
Data structures and algorithms
The core of the test. Expect arrays and strings, hash maps and sets, stacks and queues, trees and graphs, recursion, sorting and searching, two pointers, sliding windows and dynamic programming. Most problems are dressed-up versions of well-known patterns.
Correctness and edge-case handling
Hidden test cases probe empty inputs, single elements, duplicates, negative numbers, overflow and boundary conditions. Partial credit is common, so a solution that handles the awkward cases scores well above one that only works on the obvious input.
Time and space complexity
Large hidden inputs mean an inefficient solution can pass the samples and still fail. The grader effectively tests whether you reached the optimal or near-optimal complexity, so recognising when an O(n squared) approach needs to become O(n log n) or O(n) is decisive.
Computer-science fundamentals
The multiple-choice block covers data-structure properties, algorithm complexity, object-oriented concepts, databases and SQL, operating systems, concurrency and sometimes language-specific behaviour. It rewards a solid CS foundation rather than memorised trivia.
Quantitative and mathematical reasoning
A maths section covering probability, statistics, combinatorics, number theory and logic puzzles. It is weighted more heavily for quant and Strats roles, where comfort with expected value and probability is part of the job.
Coding fluency under time pressure
Translating an idea into working code quickly, in one language, without an autocomplete-heavy IDE. Knowing your chosen language well enough to write clean, bug-free code at speed is itself part of what is being measured.
The format
What to expect
See it in action
A worked example
Here is a representative medium problem in the style banks set, and the thought process that scores full marks rather than partial credit. Suppose you are given an array of integers where each value is the price of an asset on a given day, and asked for the maximum profit from buying on one day and selling on a later day.
- 01
Read the constraints first
The prompt says the array can hold up to a few hundred thousand or a million elements. That single line is the most important on the page: it tells you a solution that checks every pair of days will be far too slow and will time out on the hidden tests.
- 02
Name the brute force, then reject it
Comparing every buy day with every later sell day is two nested loops, an O(n squared) approach. It is correct and it passes the small visible samples, but on a large hidden input it exceeds the time limit and scores almost nothing. Knowing this before you type saves you the trap.
- 03
Find the linear solution
Walk through the prices once, tracking the lowest price seen so far and the best profit achievable if you sold today. Each day you update the running minimum and compare today minus that minimum against your best profit. That is a single pass, O(n) time and O(1) extra space.
- 04
Handle the edge cases
Cover the awkward inputs the hidden tests will throw at you: prices that only fall (the answer is zero, since you would not trade), a single element, an empty array, and large values that could overflow a 32-bit integer in some languages. These cases are usually where partial credit is won or lost.
- 05
Run, then submit
Test against the sample cases and a couple you invent (a falling sequence, a single element) before submitting. On submission the grader runs the full hidden suite, including a large stress test that the linear solution clears comfortably and the brute force would have failed.
The takeaway
The lesson that separates strong scores from weak ones: passing the visible samples is not the goal. The constraints tell you the complexity you need, and the hidden performance tests are what actually decide your mark. Reach for the optimal pattern before you start coding, not after a timeout.
The scoring
How it is marked
HackerRank assessments are auto-graded and produce a single composite score that firms compare against a cut-off. Coding problems are scored on the proportion of hidden test cases your solution passes, so partial credit is the norm, and the multiple-choice and maths sections add their own marks. The percentile context below is indicative; firms set their own thresholds, and the most competitive setups, such as Goldman Sachs Engineering, sit at the demanding end.
Full marks on a problem
Your solution passes every hidden test case, including the edge and large-input performance cases. This usually requires the optimal or near-optimal time complexity, not just a working answer.
Partial credit
You pass some hidden cases but fail others, typically because of an unhandled edge case or a solution that times out on large inputs. Still worth submitting, since partial scores accumulate across the assessment.
Samples only
Code that compiles and clears the visible samples but fails most hidden cases. This is the classic brute-force outcome and scores far lower than candidates expect.
Firm cut-off
Banks set a threshold on the composite score and advance only those above it. Competitive engineering and quant programmes, with Goldman among the toughest, expect strong performance across all three sections rather than a single standout problem.
The variants
Versions you might be sent
Goldman Sachs Engineering assessment
Goldman runs its Engineering coding screen on HackerRank, and its configuration is widely considered among the hardest on Wall Street: demanding algorithm problems, tight timing and a high bar across the coding, CS and maths components.
Technology coding challenge
The general software-engineering setup used across bank technology hiring, including J.P. Morgan Tech. Coding-heavy, with data-structure and algorithm problems plus CS multiple choice, and lighter on the quantitative section.
Quant and Strats variant
Used for quantitative and Strats roles, such as Morgan Stanley Quant. The coding is still central, but the probability, statistics and maths content carries noticeably more weight, reflecting the role.
CodePair live interview
HackerRank's live, shared-editor interview format. Some firms use it for a follow-up round where you code while an interviewer watches, so explaining your reasoning out loud matters as much as the solution.
Who uses it
Firms that screen with this test
Each links to a dedicated firm guide: the application process, the interview stages, and what they look for.
The prep
How to prepare
Drill data structures and algorithms
This is where most marks are won. Work through the core patterns until they are automatic: hash maps, two pointers, sliding window, stacks and queues, binary search, trees and graphs, recursion and dynamic programming. Most problems are a variation on a pattern you can learn.
Optimise, do not just solve
Train yourself to read the constraints and infer the required complexity before coding. Practise turning an O(n squared) idea into O(n log n) or O(n), because the hidden performance tests are what fail unprepared candidates who only check the samples.
Practise in the real editor, against the clock
Code in a plain browser editor without your usual IDE crutches, and sit problems under the real time pressure. Comfort with running against sample cases, finding bugs by reading, and budgeting time across two or three problems only comes from reps.
Master one language deeply
Pick one supported language (Python, Java and C++ are popular choices) and know its standard library cold: the collection types, sorting, hashing and string handling. Fighting your language wastes minutes you do not have.
Refresh CS fundamentals and quant
Revise complexity analysis, data-structure properties, object-oriented concepts, SQL and operating-system basics for the multiple choice, and brush up probability, statistics and combinatorics for the maths section, especially if you are targeting quant or Strats.
Sit full timed mock assessments
Do complete two-to-three-problem sets plus multiple choice in one sitting, so the 90 to 120 minute format is familiar. Reviewing every solution you could not optimise teaches more than simply doing more easy problems.
Practise on the real format
Reading about the test is not practising it.
Intervyo recreates HackerRank Coding Assessment in its real format, timed and scored, with instant feedback so the structure is familiar before it counts. Start free, no card required.
FAQ
Common questions
It is a timed online coding test, usually 90 to 120 minutes, used as an early screen for engineering, technology and quant roles. You solve two to three algorithm problems in a browser editor, answer computer-science multiple choice, and work through a maths section, all auto-graded against hidden test cases.
Keep exploring
Keep going
Sectors
Practise on Intervyo
HackerRank Coding Assessment
Know the test.
Now practise it on the real format.
Intervyo recreates the timed pressure of these assessments and scores every run, so the format is second nature before the real one.
Free to start, no card required
Intervyo is an independent preparation platform. Our practice tests simulate the style and format of HackerRank assessments and are not affiliated with, or endorsed by, the test publisher.