A sandboxed coding round. The code really runs against hidden test cases and is scored on correctness, efficiency and code quality. Every submission comes back with a result per case, a runtime in milliseconds and a complexity verdict.
30-day free trial with 75 Hire Credits. Card required, nothing charged for 30 days.
The engineering screen
A keyword tells you a word appeared on a page. A take-home tells you something was submitted. Neither tells you whether this person can turn a problem into code that runs on input they never saw.
Engineering loops are the most expensive rounds you run, and the two filters sitting in front of them are the least reliable evidence you have.
A resume filter matches strings. The word Python covers a weekend bootcamp exercise and eight years of production services, and the filter cannot tell them apart.
An unsupervised multi-hour project is done at a kitchen table, with an assistant open in the next tab, and arrives as a zip file with no record of how it was produced.
A reviewer skims a submission, forms an impression and writes two lines of feedback. The next reviewer applies a different bar to the next candidate on a busier day.
A six-hour take-home asks for unpaid work before you have offered anything. People who already have offers in hand simply say no and move on.
Here is what replaces both of them.
The candidate opens a link and lands in the round. Nothing to download, no environment to configure, no submission to package up. The same engine runs the proctored hiring round and the practice surface, so what a student rehearses on is what an employer runs.
Given a set of coin denominations and a target amount, return the minimum number of coins needed to make exactly that amount. Unlimited supply of each coin. If impossible, output -1.
First tab: the proctored hiring round. A Hard Coin Change problem tagged dynamic-programming, a Python editor with custom stdin, Run and Submit, a 5 of 5 passed panel with a runtime on every case, and a complexity read of O(n·k) marked optimal. Second tab: the practice surface a student sees, with Problems, Paths and My Results, an Easy Binary Search problem, and a 3 / 4 passed result naming the case that failed.
A statement, a worked example, a difficulty tag and topic tags. The screens above show a Hard dynamic-programming problem on the hiring side and an Easy Binary Search problem on the practice side.
A line-numbered editor with a language picker above it. Nothing to install, no repository to clone, no archive to email back at the end.
A custom stdin box lets the candidate paste an input and hit Run before submitting. That is how people debug, so the round allows it.
Submit executes the code against the hidden test cases. Each case comes back pass or fail with its runtime in milliseconds, above an X of Y passed counter.
Three dimensions, each backed by something you can point at. The result is a record of what the program did when it ran.
Hidden test cases decide this, not whoever happened to review the file. Every case is listed by name with a pass or fail mark, so you can see which input broke the solution.
Runtime is recorded per test case in milliseconds. A solution that clears the samples and then crawls on the large input shows up in the results panel, not six months later.
The submission is read as code, not only as output. Structure and readability are scored alongside whether the answer came out right, because someone else has to maintain it.
The hidden test cases are not string comparisons against a stored answer key. The candidate's program runs in a Piston-backed sandbox and is judged on what it produced for inputs they never saw. That is why a runtime in milliseconds exists at all. Something ran.
A candidate can hard-code the sample answers. They cannot hard-code a test case they were never shown.
Under the test results the round prints a complexity read on the submitted solution and whether it is optimal for the problem. It separates a candidate who brute-forced five green ticks from one who reached for the right algorithm.
Two submissions can both pass every case and still be very different hires. This is the line that tells you which one you are looking at.
Every score traces back to the submitted code and the cases it ran against, and it is measured against the skills and weightage you set on the job, the same as every other round. Set skills and weightage
A coding round is the easiest stage to hand to somebody else, so proctoring can be switched on for it. Expert Hire detects known AI-assistant overlay tools, signs of impersonation and tab switching during the round. Every signal goes to a human reviewer next to the submitted code.
The header of a proctored round shows a Proctored banner with the running switch count next to it. The candidate is told what is being watched, and the same number reaches the reviewer.
Browser proctoring covers lower-stakes coding rounds. A cross-platform desktop app runs strict-proctored rounds, bundling video, the code editor and security monitoring in one place.
Integrity signals sit next to the code and the transcript on the report, for a human to read. A flag is never an automatic verdict, and an integrity flag never rejects anyone.
Add it to a coding test and the round costs exactly the same. The desktop proctoring app unlocks at Growth and Scale.
A pipeline is built out of rounds, and every round is one of five types. You set each one up in Basic Details, Round Configuration, Questions and Workflow, four steps when the job has workflow on and two when it is off. Rounds are numbered inside the job, and AI and human rounds mix in the same pipeline.
AI analyses skills, experience and education against the role to produce a readiness score. No interview required.
Ethan runs a structured voice round. The type is one of eight, from General Interview to System Design, and the follow-up questions adapt to the answer.
A live meeting hosted here or on Zoom, Meet or Teams with a notetaker in the room. Scored with a full report.
Candidates write prompts for real tasks, scored on 8 dimensions including prompt quality and iteration strategy.
A sandboxed problem with real execution against hidden test cases. Scored on correctness, efficiency and code quality.
All four rounds are scored against the same skills and weightage you set on the job, so the shortlist at the end is a comparison. A workflow can auto-shortlist or auto-reject against a promotion threshold you choose, and both are off until you turn them on. Every stage lands on one share-ready report per candidate, with the recording, the searchable transcript and the submitted code behind it.
Coding tests exist twice: as a hiring round a recruiter adds to a pipeline, and as a practice surface a candidate works through alone. The engine and the scoring are the same, so a student who practiced here is not surprised by the employer version.
A recruiter adds a Coding Test round to the pipeline and candidates get it like any other stage. The result lands on the candidate report next to the interview scores, the recording and the transcript, so a debrief is one link rather than four tabs.
Job seekers get their own Coding Tests surface with a problem catalogue, guided paths and a results history. Each attempt returns the same per-case pass or fail summary an employer sees, for example 3 / 4 passed.
Placement cells get coding tests inside the campus per-student rate, with a lower rate for larger cohorts. Results roll into the cohort readiness dashboard, so the cell can see who is below the bar while there is still time to fix it.
A coding test draws 3 Hire Credits from one shared monthly pool for the whole workspace. Team members are unlimited on every tier and never charged per seat. Auto-scored coding tests are on every plan, proctoring costs nothing extra per session with the desktop app unlocking at Growth, and reports are always 0.
| Action | Credits |
|---|---|
| AI resume screen | 1 |
| Human interview + AI copilot | 2 |
| Calendar notetaker bot | 2 |
| Coding test | 3 |
| AI-prompt assessment | 3 |
| AI interview | 5 |
| Candidate matching | 0 |
| Proctoring (desktop app) | 0 |
| AI reports and insights | 0 |
The 30-day free trial comes with 75 Hire Credits and charges nothing for 30 days. At 3 credits a test that is 25 full coding rounds. Unused plan credits roll over for one month, and top-up packs stay valid for three.
Job seekers buy Prep Credits instead, in one-time packs with no subscription, where a practice coding test costs 3. Universities license coding tests per student as one module of the placement suite.
See plans and creditsWhat engineering leads and recruiters ask before they add the round.
Still have a question? Book a demo and we will run a coding round with you live.
Add a Coding Test round to your next engineering job and see hidden test cases, per-case runtimes and a complexity verdict on every candidate. 3 Hire Credits a test, proctoring optional and free, reports never metered.
30-day free trial with 75 Hire Credits. Card required, nothing charged for 30 days.