The Amazon Online Assessment Is Four Tests, Not One
You can finish both coding problems on the Amazon online assessment with fifteen minutes to spare and still get a rejection a week later. It’s one of the most common questions we get from candidates prepping on LastRound AI: “My code passed every test case, so why didn’t I move to the phone screen?” The answer is almost always sitting in the three sections that have nothing to do with writing code.
Amazon’s own careers page lays the whole thing out. For a full-time SDE role the online assessment runs close to two hours across four timed parts: a coding assessment (around 70 minutes), a Work Styles assessment (about 15), a Work Simulation (about 60), and a short feedback survey (about 5), per Amazon’s official SDE online assessment guide. Only one of those four is about code. Most prep material acts like the other three don’t exist.
What the Amazon online assessment actually tests
The Amazon online assessment is one timed session that scores two very different things: how well you write code, and how closely your stated working style lines up with Amazon’s 16 Leadership Principles. It isn’t a pure coding screen. Candidates usually take it remotely on a proctored coding platform (most report HackerRank) after a recruiter sends an invite from amazon.jobs, and the exact mix of sections depends on the role you applied for.
Two things are worth knowing before you start. The invite email tells you which sections your version contains, so read it instead of assuming. And Amazon sends a status email as you finish each section, so you sometimes learn whether you’ve advanced partway through rather than all at once.
The four sections, and how long each one takes
Here’s the full-time SDE breakdown, with the internship version noted where it differs. The internship online assessment drops the Work Simulation and comes in around 90 minutes total.
| Section | What it measures | Typical time |
|---|---|---|
| Coding assessment | Two DSA problems, written and run in the browser | ~70 min (Intern / SDE I), ~90 min (SDE II) |
| Work Simulation | Workplace scenarios: prioritising and responding, no code | ~60 min (full-time SDE; not in the internship OA) |
| Work Styles Assessment | Forced-choice statement pairs mapped to the Leadership Principles | ~15 min |
| Feedback survey | Not scored. Your feedback on the assessment itself. | ~5 min |
SDE II candidates get a different shape, which I’ll come back to below.
The coding section is the part everyone over-prepares
Two problems. That’s it. They sit around LeetCode medium, occasionally nudging into hard, and they’re graded on more than whether the visible tests go green. JobTestPrep’s 2026 breakdown lists the scoring dimensions as correctness, efficiency, data-structure choice, readability, and edge cases, in their Amazon OA guide, updated June 2026. A brute-force solution that clears the sample cases can still lose points on the hidden ones and on time complexity.
Two practical things candidates report, and that we’ve seen echoed. Amazon expects syntactically correct code written in real time, and the platform flags large paste events. So the person who memorised fifty solutions and plans to paste from a second monitor tends to do worse, not better, than the person who can write one clean solution slowly.
To give a sense of the level (this is illustrative of the pattern, not a real leaked question): you might get an array of package weights and a number of days, and be asked for the minimum daily ship capacity that clears everything on time. That’s a binary-search-on-the-answer problem wearing Amazon-flavoured logistics language. Recognise the pattern under the story and you’re most of the way there.
The two sections that quietly decide the result
This is where strong engineers lose the Amazon online assessment. The Work Simulation and the Work Styles assessment don’t test code at all. They test whether your instincts match how Amazon says it wants people to work, and a candidate who aces the coding can still get filtered here without ever seeing why.
The Work Simulation drops you into a fake Amazon workday. Emails and instant messages arrive from a virtual team lead and a couple of colleagues, a problem is unfolding, and you choose how to respond and what to prioritise. Nothing on screen is labelled right or wrong, but there’s a scoring key underneath, and it rewards responses that read as Customer Obsession, Ownership, and Bias for Action.
The Work Styles assessment is blunter. You get pairs of statements (“I prefer clearly structured tasks” versus “I look forward to learning new things”) and pick which one fits you better. Every pair maps back to the Leadership Principles. Amazon lists 16 of them now, including the two added in 2021, Strive to be Earth’s Best Employer and Success and Scale Bring Broad Responsibility, on its official Leadership Principles page. Honestly, I don’t rate this section much as a hiring signal. Forced-choice personality items are easy to game and easy to overthink at the same time. You still have to pass it, so answer as the version of yourself who has already internalised those 16 principles, not the version reading them for the first time mid-test.
One caveat worth stating plainly: Amazon doesn’t publish a passing bar for the Work Styles section, and I haven’t seen a trustworthy one anywhere, so treat any blog quoting you an exact cutoff as guessing.
SDE II swaps the simulation for system design
At the SDE II (L5) level the format shifts. The coding window stretches to about 90 minutes, the Work Simulation drops away, and a multiple-choice System Design section takes its place, roughly 15 to 20 minutes on architectural trade-offs rather than a live whiteboard, again per JobTestPrep’s 2026 guide. The Work Styles assessment stays. It’s a reminder that even at the senior level, the Leadership Principles aren’t something you clear later in the loop. They’re baked into the very first automated filter.
How to prep for the parts that aren’t LeetCode
Grind the two coding problems until they bore you, sure. For most candidates, though, the higher-return hour is the one spent on the Leadership Principles, because that’s the same muscle the Work Styles survey, the Work Simulation, and every later behavioural round all pull from. The trap is treating them as trivia to skim. They’re the rubric your answers get scored against, in the OA and in the onsite loop.
The way we’d build that muscle: rehearse a few real stories out loud, each mapped to a specific principle, until the framing feels automatic. That’s the point of LastRound AI’s voice mock interviews, which run a live AI interviewer and score each answer on six dimensions (relevance, content, structure, specificity, clarity, and conciseness) plus a filler-word count. The survey itself is multiple choice, so you’re not literally rehearsing it, but the consistency carries over. Someone who can already tell an Ownership story cleanly tends to answer the Work Styles pairs consistently too, because they actually know which principle they lead with.
For the coding side and the behavioural questions that come after the OA, our breakdown of the Amazon SDE-2 interview loop covers the specific problems and Leadership-Principle questions candidates report from the phone screen and onsite.
Almost nobody practises the coding part
One number from our own data is worth putting next to the four-test breakdown above.
Of 1,393 interview sessions configured on LastRound between January 2025 and July 2026, exactly 8 had the coding round switched on. The other 1,385 didn’t.
Eight. Out of 1,393.
Amazon’s assessment is weighted heavily toward the two coding problems in Test 1, and it’s the part candidates say they fear most. Yet when people sit down to practise with a tool that offers a coding mode, they practise talking instead.
I have a guess about why. Talking practice is comfortable and coding practice feels like work you could be doing on LeetCode instead. I can’t prove that from the data, and it might just be that our coding mode isn’t discoverable enough.
FAQ
How long is the Amazon online assessment?
The internship online assessment runs about 90 minutes, and the full-time SDE version runs up to two hours, because it adds a roughly 60-minute Work Simulation on top of the coding and Work Styles sections. Amazon’s careers page gives the per-section averages, and your invite email confirms which sections your specific version includes.
How many coding questions are on the Amazon OA?
Two. Both sit around LeetCode medium difficulty, occasionally harder, and you get roughly 70 minutes for the pair at the intern and SDE I level, or about 90 minutes at SDE II. They’re scored on correctness, efficiency, and code quality, not just whether the visible test cases pass.
Can you fail the Amazon online assessment even if your code works?
Yes, and it’s common. The Work Simulation and Work Styles assessment are scored against Amazon’s Leadership Principles, and a candidate whose responses read as inconsistent or misaligned can be rejected despite a perfect coding score. Treat those sections as real, gradeable parts of the test rather than an afterthought.
Is there a debugging section on the Amazon OA in 2026?
Amazon’s current official rundown of the SDE assessment doesn’t list a standalone debugging section, and the 2026 format we verified is coding, Work Simulation (full-time), Work Styles, and a survey. Older versions and some hiring batches did include a debugging round, so this is exactly the kind of detail worth confirming in your own invite email rather than trusting any single guide, including this one.
Practise the two coding problems until they’re dull. Then spend one honest evening deciding which of Amazon’s 16 Leadership Principles you can actually back with a real story. That second hour is the one most candidates skip, and it’s the one the assessment is quietly grading hardest.
Written by
Shekhar Babu
Writes about technical interview rounds, system design and coding assessments.
