A timed coding round built to match the online assessments (OAs) companies actually send you. You choose the format and the language, you solve against the clock, and at the end an AI reviewer explains not just what you got wrong, but where your time went and which problem you should have opened first.
4
Assessment formats
4
Languages
45 to 90
Minutes per round
def longest_window(events, limit):
left, best, seen = 0, 0, {}
for right, e in enumerate(events):
seen[e] = seen.get(e, 0) + 1
...
Each format below copies the problem count, difficulty mix and time limit of an online assessment that companies genuinely run. Pick the one that matches the employers you are applying to.
2 problems
Easy
Medium
Two problems in 45 minutes: one easy, one medium. This is the first-filter format most employers use. The easy problem is meant to be cleared quickly, so the real test is whether you leave yourself enough time for the medium one.
2 problems
Medium
Medium
Two medium problems in 70 minutes, with no easy warm-up. Some companies skip the gentle opener, so both problems carry equal weight and your pacing decides whether you finish.
4 problems
Easy
Medium
Medium
Hard
Four problems in 70 minutes, rising from easy to hard. Most people do not finish all four, and that is deliberate. Choosing what to attempt and what to leave is part of what the review coaches you on.
4 problems
Easy
Medium
Medium
Hard
The same four problems as the sprint, with 90 minutes instead of 70. There is room to think properly, so you have time to write a hard-problem solution that scales instead of one that runs past the time limit on the larger tests.
Start by choosing the language you want to code in. Then pick one of the four assessment formats: two short screens of two problems, and two longer rounds of four. Each format sets its own time limit and difficulty mix, and each one is modelled on an online assessment that companies genuinely send out.
You get a full-screen code editor, a countdown timer, and your list of problems down the side so you can move between them freely. Run your code against the visible test cases as often as you like while you work. When you submit a problem it is locked in, and a second set of hidden tests runs against it.
When the round ends, an AI reviewer reads your code, your timings, and the order you worked through the problems. You get written feedback on each problem, plus a summary of how you spent your time across the whole round and what you should have done differently.
Every round ends with the share of tests you passed, per-problem coaching, and notes on your approach, pacing, complexity and code quality. Here is what counts toward the score and what is there to coach you.
Your score is the share of tests you pass, visible and hidden, the way a real online assessment reports it. The edge-case, ordinary and medium-input groups count equally for correctness, and the largest inputs are the performance half. A test that runs past the time limit for your language counts as failed, so a brute force that times out scores below a solution that scales.
How you attacked each problem, read from the code you actually wrote. Coaching on the idea behind your solution, not a number.
How you spent your minutes against the format you chose. A note, not a score.
Whether your approach scaled, read from which tests finished inside the time limit. You never state it yourself.
Naming, structure, dead code. Feedback on how it reads, kept out of the number.
The round records when you opened each problem, how long you stayed on it, and whether you gave up and came back later. Most people lose an assessment by spending forty minutes on the wrong problem, so the review shows you where your minutes actually went and which order would have served you better. It is coaching, not part of your score.
The visible test cases are there for you to work against while you code. A second set of hidden tests runs only when you submit, and those results stay hidden until the round ends, so you cannot reverse-engineer the answer from the feedback.
Every round excludes every problem you have seen in your last three rounds. Retaking a format gives you a genuinely new set, so your score reflects your ability rather than your memory.
Scores never mix between formats. A 45-minute two-problem screen and a 90-minute four-problem round test different skills, so each keeps its own history and its own progress chart. Improving at short screens will not inflate your full-round score.
Each problem ships with a reference solution that has been machine-run and confirmed to pass in your chosen language before that problem can be served. You will never lose time to a broken problem or a test case that cannot be satisfied.
When the time runs out, whatever you have is collected and scored, exactly as a real online assessment does. There is no pause button and no overtime, because there is not one on the day either.
From there, every round you run is measured against that first one. Your scores and your pacing are tracked over time, and each format keeps its own history so you can see exactly where you are improving and where you are not.
The DSA round tests whether you can solve problems correctly on your own under time pressure. The mock technical interview tests something different: whether you can explain your thinking out loud to someone who keeps asking why. Six stages, an AI interviewer that follows up on your answers, and a written report on how you reasoned rather than just what you produced.
Six stages covering technical questions, live coding, your own project, system design, code review and building with an AI assistant.
Run a single stage to drill one weakness, or stack several back-to-back and get one combined report for the whole sitting.
Scored on communication and depth under probing, not only on whether your code ran.
It is the timed coding test most companies send you before anyone speaks to you. You get a small number of data-structures-and-algorithms problems, a fixed time limit, and hidden test cases you cannot see while you work. It is usually the first filter in the hiring process, and it is where most candidates are rejected.
There are four formats. Two short screens of two problems each, at 45 and 70 minutes. Two longer rounds of four problems, at 70 and 90 minutes. The four-problem rounds always run easy, medium, medium, hard, so the only thing you choose there is how much time you get.
Python, Java, C++ and JavaScript. You pick your language when you build the round, and every problem has been verified to be solvable in the language you chose before it is served to you.
Each problem comes with visible test cases you can run as often as you like while you code, plus a second set of hidden tests you never see. The hidden tests only run when you submit, and their results stay sealed until the round finishes. Real online assessments work exactly this way, which is why a solution that passes the examples can still fail the assessment.
Each problem is scored on the share of its tests you pass, the way real online assessments do it, in two halves: correctness, from the groups of tests that check edge cases, ordinary inputs and medium-sized inputs, which count equally, and performance, from the share of the largest-input tests your solution passes inside the time limit. A test that runs past the time limit counts as failed, and a problem left as the starter code scores zero. Your round score is the average over the problems. Your approach, pacing and code quality are reported as coaching notes and are not part of the score.
No, and on the tighter formats you probably will not. That is deliberate. Deciding which problem to attempt, which to leave, and when to give up on one and move to the next is a real skill. The review tells you where your minutes went and which order would have served you better, as coaching rather than as part of your score.
Yes, and you will not be handed the same problems again. Every round excludes every problem you have seen in your last three rounds, so a retake is a genuine retake rather than a memory test.
Whatever you have is collected and scored, exactly as a real online assessment does. There is no pause button and no overtime, because there is not one on the day either.
Solving problems one at a time with no clock trains a different skill from sitting a timed assessment. Here you get the whole round: a fixed time budget across several problems, hidden tests, and no ability to look up the answer. You also get coaching on your pacing and on whether your solutions scale inside the time limit, which is where most candidates actually lose the assessment.
No. Each format keeps its own history and its own progress chart, because a 45-minute two-problem screen and a 90-minute four-problem round test genuinely different skills. Getting better at short screens will not inflate your score on the longer rounds.