Courses › Coding Interviews › Getting Started

Online Assessments

Timed, auto-graded tests: formats, hidden tests, partial credit, pacing and integrity rules.

Many companies start the coding loop with an online assessment (OA): a timed, automatically graded test you take alone, usually on HackerRank, CodeSignal, Codility or a company's own platform. Nobody watches you think, and nobody gives hints. Only the test results count, so OAs reward a different set of habits from live interviews.

Common formats

Read the invitation email carefully. It usually names the platform, the time limit, the number of questions and whether you can use your own IDE or search for documentation.

How scoring works

Hidden test cases decide your score. The visible examples are only a small sample. Hidden tests cover edge cases and large inputs, and a solution that's too slow fails the big ones with a time limit error.
Partial credit usually exists. Passing 12 of 15 tests often counts for something. A brute-force solution that passes the small tests is worth far more than an elegant one that doesn't compile.
Some platforms record more than results: your code, the time taken, keystroke playback, and sometimes webcam or screen recording. Reviewers may read your code if your score is borderline, so keep it readable.

[!WARNING] Integrity rules are strict Unless the instructions say otherwise, assume no AI assistants, no help from others and no copying from the web. Many platforms detect pasted code, tab switches and second devices, and a flagged result can block you from reapplying for a long time. When documentation lookups are allowed, the instructions say so.

A pacing strategy

Skim every problem first (2–3 minutes). Difficulty isn't always in order, and on leveled OAs the later problems are often worth more points.
Bank the easy points. Solve the easiest problem quickly and completely, then submit it.
Get something passing on every problem. A brute force that passes small tests beats a blank. Then optimize the one with the most points left.
Check the constraints before choosing an approach. They tell you the target complexity (see the table below).
Leave 10 minutes at the end to re-run everything and check for leftover debug prints. Extra output can make correct answers fail.

Reading the constraints

Testing without hints

On an OA you're your own interviewer. Before submitting, run your code on cases you write yourself:

Empty and single-element inputs
All elements equal; already sorted; reverse sorted
Negative numbers, zeros and the largest allowed values (watch for overflow in Java and C++)
The largest input size, to check speed. A quick generator helps: [random.randint(1, 109) for _ in range(105)]

Input, output and platform details

Function stubs vs standard input. Most OAs give you a function to fill in. Some make you read from standard input yourself. In Python, read everything at once with sys.stdin.read().split() for speed.
Know your language's limits. In Python, deep recursion fails at about 1,000 levels by default; use an explicit stack or raise the limit (see Python for Interviews).
Don't rely on autocomplete. Some platforms have a basic editor. Practice writing code without IDE help.
Your environment. A quiet room, a stable connection, a charged laptop, and any required ID ready. Close other apps if the test is proctored.

[!TIP] Practice in the real format Most platforms offer a sample test. Take it a few days before, to learn the editor, how input works and how submissions are graded. Then do two or three timed sets of 2–3 mixed problems from the question library.

After the OA

Results usually arrive within a few days to two weeks. A pass rarely gives you a score, and a reject rarely explains why. Either way, write down the problems while you remember them: they show which patterns to review, and similar ones often come up again at the next company.