Amazon Online Assessment Prep Checklist
For many software roles, the Amazon process starts with an online assessment (OA) link instead of a person. Nobody is there to ask questions, give hints, or tell you you're on the right track. It's you, a timer, and a set of hidden test cases. Many candidates find that harder than a live round, because there's no one to read and no feedback until the end.
This Amazon online assessment prep checklist covers what to practice and how to prepare for the day itself. A caveat up front: OA formats change over time and differ by role, level, and region. What follows reflects what candidates commonly report, not an official description. Read the invitation email carefully and check with your recruiter about the sections, the time limits, and the allowed languages.
Step 1: confirm the format
Before you plan your practice, write down the answers to these from the invite or your recruiter:
- [ ] How many sections are there, and what is each one (coding, work-style survey, work simulation, written explanation)?
- [ ] Time limit per section, and whether you can pause between sections
- [ ] Which languages are allowed
- [ ] Whether the platform shows the results of hidden tests while you work
- [ ] Deadline for completing the assessment
- [ ] Whether it's proctored or recorded, and any rules about the environment
Candidates often describe a coding section with a small number of problems under one shared timer, plus a work-style or behavioral section linked to Amazon's Leadership Principles. Confirm the details for your role instead of assuming.
Step 2: practice under OA conditions
OA practice is different from live-interview practice. There's no interviewer to talk to, so your focus shifts to:
- Reading carefully. Problem statements can be long and wrapped in a story. Pull out the real input, the output, and the constraints.
- Hidden tests. Your code has to handle edge cases you won't see.
- Pacing across problems. One shared timer means you choose how to split it.
- No outside help. Practice the way the OA rules require, with no searching and no assistance.
Build sessions that match:
| Session element | How to practice |
|---|---|
| Timer | Same total time as your OA, all problems on one clock |
| Environment | An online judge or plain editor in your chosen language |
| Problem mix | One easier and one medium-to-harder problem, unseen |
| Rules | No searching, no notes beyond what the rules allow |
| Review | Afterwards, list which edge cases you missed |
Step 3: cover the patterns that show up often
Candidates commonly report array and string problems, hash maps, sorting, greedy choices, heaps, and graph or grid traversal. A focused list:
- Hash maps and frequency counting
- Sorting plus a greedy pass (scheduling, assigning, minimizing cost)
- Two pointers and sliding window
- Heaps for "k smallest" or "repeatedly take the largest"
- BFS and DFS on grids (islands, shortest path in a maze)
- Prefix sums for range queries
- Binary search on the answer ("smallest capacity that works")
If you only have a week, prioritize hash maps, sorting plus greedy, heaps, and grid BFS.
Step 4: read constraints like a scoring guide
In an OA, the constraints tell you how much of the hidden test suite a brute force is likely to pass.
| n up to | Brute force likely OK? |
|---|---|
| ~1,000 | O(n²) often passes |
| ~100,000 | Need O(n log n) or O(n) |
| ~1,000,000+ | O(n) with small constant factors |
Rule of thumb: if the constraints are large and you only have a quadratic idea, code it anyway, submit it to bank partial credit (if your platform gives partial credit), then optimize.
Step 5: a pacing plan for one shared timer
Assume two problems on one clock:
- Read both problems first (about 3–5 minutes). Decide which one looks easier.
- Solve the easier one first, fully, with edge cases.
- Submit, or save, a working version before you start optimizing.
- Move to the harder one and get a brute force working before anything clever.
- Use the remaining time to optimize whichever problem has the most failing tests.
- In the last 5 minutes, stop starting new ideas. Make sure what's saved compiles and runs.
An accepted easier problem plus partial credit on the harder one is a common, respectable outcome.
Step 6: an edge-case checklist for hidden tests
Before you submit each problem, check:
- [ ] Empty input, or the minimum size the constraints allow
- [ ] Single element
- [ ] All elements equal
- [ ] Already sorted, and reverse sorted
- [ ] Very large values (could the sum overflow 32-bit integers in your language?)
- [ ] Negative numbers and zero, if allowed
- [ ] Ties: does the problem say which answer to return?
- [ ] Output format exactly as specified (order, type, spacing)
Overflow is easy to miss in Java and C++. If the values can reach about 10⁹ and you add them up, use long or long long.
Step 7: prepare for the work-style section
If your OA includes a work-style survey or work simulation, don't wing it. Read Amazon's publicly published Leadership Principles and think about how you actually work. Answer honestly and consistently. These sections tend to present realistic scenarios, so it helps to reflect on real situations from your own experience: a time you disagreed with a teammate, prioritized under a deadline, or dug into a customer problem.
There's no need to memorize "right answers." Inconsistent, performative answers are more likely to hurt you than help.
Step 8: set up the day
- [ ] Quiet room, stable internet, laptop plugged in
- [ ] Browser updated; close extra tabs and notifications
- [ ] Platform tested if they offer a practice question
- [ ] Water nearby, phone silenced and out of reach
- [ ] Started well before the deadline, not in the last hour
- [ ] Language and editor settings you're comfortable with
Common OA mistakes
- Starting with the harder problem and running out of time
- Optimizing before anything works
- Not reading the output format and failing on formatting
- Ignoring overflow in typed languages
- Rushing the work-style section as if it doesn't count
- Taking the OA tired, late at night, just before the deadline
Soft practice note
The OA rewards habits you build beforehand: careful reading, a pacing plan, and a quick edge-case pass, all under a real timer. If you'd like structured timed practice sessions to build those habits and stay calm with a clock running, Luna Interview is designed for prep and rehearsal. Take the assessment itself by its rules.
Final checklist
- [ ] Format, timers, and languages confirmed
- [ ] Three or more full timed practice sessions done
- [ ] Core patterns reviewed (hash maps, greedy, heaps, grid BFS)
- [ ] Pacing plan: read both, easier first, save working versions
- [ ] Edge-case and overflow checklist memorized
- [ ] Leadership Principles read and real examples reflected on
- [ ] Environment and timing set up for the day
An OA is a fair test of preparation. Practice it the way you'll take it, and the timer becomes something you plan around.
Rehearse with an AI interview copilot, then work through commonly reported Amazon interview questions.