Amazon SDE Online Assessment 2026: Questions, Format & Prep
What the Amazon and AWS software development engineer online assessment looks like for interns, new grads and experienced engineers, the coding patterns it tests, sample problems with solutions, and how to prepare.
What Is the Amazon SDE Online Assessment?
The Amazon SDE online assessment (OA) is the first screening step for most software development engineer roles at Amazon and AWS. It is a timed, remote test that usually contains two coding problems followed by a Work Style survey based on Amazon's Leadership Principles. You complete it on your own computer, normally within about a week of receiving the invitation, and your result decides whether you move on to interviews.
The OA matters more at Amazon than at many other large tech companies because Amazon hires software engineers at very high volume. For university hiring in particular, the OA filters a large applicant pool down to the candidates who are invited to the interview loop. For experienced hires, it often replaces or precedes a technical phone screen.
This guide is specific to software engineering. If you applied for a warehouse, operations, finance or corporate business role, the assessment is different: see our Amazon online assessment guide for the Work Simulation and aptitude tests used for those roles, and the Amazon aptitude test overview for the full picture across job families.
Typically two coding problems (medium difficulty, often in a delivery or warehouse setting) in roughly 70–105 minutes, then a Work Style survey of statements about how you work. Some candidates also report a short work simulation, and experienced SDE II candidates may see a system design component. Your invitation email is the authority on your exact format.
Amazon does not publish an official OA specification, and the structure has changed several times. Older guides describe three separate stages (a debugging section, a coding section and a work simulation). Most 2025–2026 candidate reports describe a combined coding plus Work Style assessment. Everything below is based on Amazon's public hiring guidance and consistent candidate reports, so treat timings as typical rather than guaranteed.
Amazon OA Format by Role: Intern, SDE I, SDE II and AWS
The core of every SDE assessment is the same: solve coding problems under a time limit, then describe your work style. What changes by level is the number of extra components and how much the OA counts relative to later interviews.
| Role | Typical OA components (candidate-reported) | What comes next |
|---|---|---|
| SDE Intern | 2 coding problems; Work Style survey; sometimes a short work simulation | Often one or two technical interviews, or a short final round. Some cycles have moved strong OA candidates to offers with very little live interviewing. |
| SDE I (new grad) | 2 coding problems; Work Style survey; sometimes a work simulation | Virtual onsite loop of about 3 interviews, each mixing coding and Leadership Principle questions |
| SDE II (experienced) | 2 coding problems; Work Style survey; some candidates report a system design or design-reasoning section | Phone screen in some cases, then a loop of about 4–5 interviews including system design and a Bar Raiser |
| AWS SDE | Same as the matching SDE level; AWS teams use the same software engineering bar | Same loop structure, with more emphasis on distributed systems for experienced roles |
| AWS Cloud Support / technical non-SDE | Varies: candidates report technical knowledge questions (networking, Linux, troubleshooting) instead of LeetCode-style coding | Technical phone screen and loop focused on fundamentals and Leadership Principles |
A few practical details are consistent across levels. The invitation usually gives you several days to start, and once you begin a timed section you cannot pause it. You can normally choose your programming language from a list that includes Python, Java, C++, JavaScript and others. Each problem has a built-in editor with sample tests you can run, and your code is then graded against hidden tests.
Pick one language and stick with it. Python is popular because it is concise, but Java and C++ candidates do equally well if they know their standard library. Make sure you can write a heap, a deque-based queue and a sort with a custom comparator in your language without looking anything up.
If you are comparing processes, the Google interview process, Meta and Microsoft use similar coding screens, but Amazon is distinctive in how heavily it weights Leadership Principles even at the OA stage.
The Coding Section: Topics, Difficulty and Scoring
Amazon OA problems are usually medium difficulty, with the occasional easy warm-up or harder optimization question. They are almost always framed as an Amazon business story: packages and trucks, warehouse shelves, delivery routes, Prime orders, server loads or product reviews. The story is decoration. Your first job is to strip it away and recognize the underlying pattern.
Most common topics
| Pattern | How it tends to appear | Target complexity |
|---|---|---|
| Hash maps & counting | Duplicate orders, frequency of items, matching pairs | O(n) |
| Sorting + greedy | Loading trucks, scheduling deliveries, minimizing cost | O(n log n) |
| Two pointers / sliding window | Best run of consecutive days, subarrays under a limit | O(n) |
| Prefix sums | Range totals, items between markers, balancing loads | O(n) build, O(1) query |
| Heaps / priority queues | k closest locations, top-k products, merging streams | O(n log k) |
| Graphs (BFS/DFS, union-find) | Connected warehouses, shortest route in a grid, dependency order | O(V + E) |
| Stacks | Next greater element, valid sequences, skyline-style problems | O(n) |
| Dynamic programming (less common) | Counting ways, minimum cost paths | Usually O(n) or O(n·m) |
How the coding section is scored
Each problem runs against a set of hidden test cases, including large inputs designed to catch slow solutions. Candidates generally receive partial credit for the cases they pass, so a working brute-force solution is better than nothing. However, a quadratic solution on a problem that expects O(n log n) will usually time out on the biggest cases. Amazon does not publish a passing score.
Code quality matters less in the OA than in a live interview, because the grader cannot see your reasoning. Still, write clearly named variables: some candidates report that recruiters or interviewers can view OA code later, and it is a good habit for the loop.
Invitations often state that the platform may record activity such as tab switching, copy and paste and, in some cases, webcam footage. Using AI tools or outside help is against the rules, can lead to disqualification and, more practically, leaves you unprepared for the live loop where you must code in front of an interviewer.
For a platform-level overview of HackerRank, Codility and CodeSignal, including their editors and test-case behavior, see our coding assessment test guide.
Sample Amazon OA Problems With Solutions
These problems are written in the style candidates describe. They are not copied from the real assessment, but they test the same patterns at a similar difficulty. Try each one with a 35-minute timer before reading the solution.
1. Minimum delivery trips (sorting + two pointers)
A courier van can carry at most two packages per trip, and the total weight on a trip cannot exceed limit. Given a list of package weights (each no heavier than limit), return the minimum number of trips needed.
def min_trips(weights, limit):
weights.sort()
lo, hi = 0, len(weights) - 1
trips = 0
while lo <= hi:
# Pair the lightest remaining package with the heaviest if they fit
if weights[lo] + weights[hi] <= limit:
lo += 1
hi -= 1 # the heaviest package always leaves on this trip
trips += 1
return tripsWhy it works: the heaviest package must go on some trip, and the best partner for it is the lightest package. If even the lightest does not fit, nothing does. Sorting gives O(n log n) time.
2. Busiest stretch of days (sliding window)
Given daily order counts and an integer k, return the largest total number of orders in any k consecutive days.
def busiest_stretch(orders, k):
window = sum(orders[:k])
best = window
for i in range(k, len(orders)):
window += orders[i] - orders[i - k] # slide one day forward
best = max(best, window)
return bestWhy it works: each step adds the new day and removes the day that left the window, so the whole array is processed once in O(n) time instead of re-summing every window.
3. Closest delivery locations (heap)
Given a list of (x, y) customer locations and an integer k, return the k locations closest to the depot at (0, 0).
import heapq
def closest_locations(points, k):
# Max-heap of size k using negative distances
heap = []
for x, y in points:
d = x * x + y * y # no square root needed for comparison
heapq.heappush(heap, (-d, x, y))
if len(heap) > k:
heapq.heappop(heap) # drop the farthest of the k+1
return [(x, y) for _, x, y in heap]Why it works: keeping only k items in the heap gives O(n log k) time, which beats sorting all n points when k is small. Mention ties and output order if the problem specifies them.
4. Connected fulfillment zones (grid BFS)
A grid marks open warehouse cells with 1 and walls with 0. Count the number of separate zones of connected open cells (up, down, left, right).
from collections import deque
def count_zones(grid):
rows, cols = len(grid), len(grid[0])
seen = set()
zones = 0
for r in range(rows):
for c in range(cols):
if grid[r][c] == 1 and (r, c) not in seen:
zones += 1
queue = deque([(r, c)])
seen.add((r, c))
while queue:
cr, cc = queue.popleft()
for dr, dc in ((1, 0), (-1, 0), (0, 1), (0, -1)):
nr, nc = cr + dr, cc + dc
if 0 <= nr < rows and 0 <= nc < cols \
and grid[nr][nc] == 1 and (nr, nc) not in seen:
seen.add((nr, nc))
queue.append((nr, nc))
return zonesWhy it works: every open cell is visited once, so the run time is O(rows × cols). Using an iterative BFS avoids recursion-depth errors on large grids, which is a common reason recursive DFS solutions fail hidden tests in Python.
Empty input, a single element, k equal to the array length, all values identical, negative numbers and very large inputs. Before submitting, run at least one of these by hand or with a quick custom test.
Work Style Survey and Work Simulation for SDEs
After the coding section, most SDE candidates complete the Work Style survey. You see pairs or groups of statements and choose how strongly each one describes you, or which of two statements is more like you. The items map to Amazon's Leadership Principles, such as Customer Obsession, Ownership, Bias for Action, Dive Deep and Deliver Results.
Strong coders do sometimes fail at this stage. The survey checks whether your stated preferences fit how Amazon expects engineers to work: taking ownership of problems end to end, moving quickly with incomplete information when a decision is reversible, digging into data, and holding a high bar for quality.
| Statement style | Leadership Principle | What a strong profile shows |
|---|---|---|
| "I fix problems I find, even when they belong to another team." | Ownership | You act beyond your immediate task |
| "I prefer to wait until I have all the information before deciding." | Bias for Action | Calculated risk-taking on reversible decisions, not paralysis |
| "I check the details myself rather than relying on summaries." | Dive Deep | Comfort with data and root-cause analysis |
| "Meeting the deadline matters more than meeting the quality bar." | Insist on the Highest Standards | You balance speed with quality instead of trading it away |
Answer honestly and consistently. The survey repeats themes in different wording, and an unrealistic "perfect" pattern can look inconsistent. Our Amazon Work Style assessment questions post walks through more example statements, and the Amazon interview questions guide covers all 16 Leadership Principles with STAR answers.
Some SDE invitations also include a short work simulation: a series of realistic engineering scenarios, such as a production issue close to a launch date or a disagreement about a design choice, where you pick the most and least effective responses. It works like a situational judgment test. The best answers usually protect customers, use data to decide, escalate early with a proposed fix and avoid blaming others.
AWS Online Assessments: SDE, Cloud Support and Solutions Roles
"AWS online assessment" can mean several different tests, because Amazon Web Services hires for software engineering, technical support, solutions architecture and many business roles.
- AWS Software Development Engineer. AWS SDE roles use the same Amazon SDE online assessment described above: coding problems plus the Work Style survey. Experienced roles lean more heavily on distributed systems in the later loop.
- Cloud Support Associate / Engineer. Candidates commonly report an assessment and interviews focused on technical fundamentals rather than algorithm puzzles: networking (DNS, TCP/IP, HTTP, subnets), Linux and operating systems, troubleshooting scenarios, and sometimes databases or scripting. Review these fundamentals rather than grinding LeetCode.
- Solutions Architect and technical customer-facing roles. These rarely involve a LeetCode-style OA. Expect technical phone screens on AWS services and architecture, plus a strong focus on Leadership Principles and customer scenarios.
- AWS business, finance and operations roles. These follow the corporate assessment path (Work Style, work simulation and, for finance, numerical reasoning) described in our Amazon online assessment guide.
The fastest way to know which AWS assessment you face is the role title and the sections listed in the invitation email. If the email mentions "coding challenges", prepare as for the SDE OA. If it mentions technical knowledge or troubleshooting, focus on fundamentals.
After the OA: The Loop, the Bar Raiser and the Timeline
Amazon does not send detailed OA scores. If you pass, a recruiter usually contacts you, often within one to three weeks, to schedule interviews. If you do not pass, you may receive a rejection email or simply hear nothing for a while.
Online assessment
Coding problems plus the Work Style survey, completed remotely.
Phone screen (some roles)
More common for experienced hires: a live coding problem in a shared editor plus one or two Leadership Principle questions.
Interview loop
Usually around 3 interviews for new grads and 4–5 for experienced hires, each about an hour. Expect data structures and algorithms, object-oriented or low-level design, and system design for SDE II and above.
Debrief and decision
Interviewers submit written feedback, then discuss together. One interviewer is often a Bar Raiser, a trained interviewer from outside the team whose job is to keep the hiring standard high.
Every loop interview typically opens with one or two behavioral questions tied to Leadership Principles, so it is worth preparing six to eight stories in the STAR format before you even take the OA. Our software engineer interview questions guide covers the technical side of the loop, and the behavioral interview questions guide helps you build the stories.
4-Week Preparation Plan for the Amazon SDE OA
Most candidates receive the OA within days or weeks of applying, so start preparing when you apply rather than when the email arrives.
| Week | Focus | Target output |
|---|---|---|
| Week 1 | Arrays, strings, hash maps, sorting, two pointers | 15–20 easy and medium problems; review standard library syntax |
| Week 2 | Sliding window, prefix sums, stacks, heaps | 15 medium problems; time each one and note your average |
| Week 3 | Graphs (BFS, DFS, union-find), intervals, basic dynamic programming | 12–15 medium problems; practice spotting the pattern behind story-style prompts |
| Week 4 | Full timed mocks and Leadership Principles | 3 mock OAs (2 problems in 70–90 minutes); write 6–8 STAR stories |
Test-day tactics
- Read both problems first. Start with the one you recognize, bank the points, then spend the remaining time on the harder one.
- State the constraints. Input sizes tell you the expected complexity: around 105 elements means you need roughly O(n log n) or better.
- Get a correct solution, then optimize. A passing brute-force solution earns partial credit if you run out of time on the optimal one.
- Test custom inputs. Run empty and single-element cases before you submit.
- Use a quiet room and a stable connection. Close other applications and keep your phone away to avoid proctoring flags.
For general online test strategy, including how employers use assessment results, see our online assessment test guide. If you are applying for an internship, our internship interview guide covers the shorter intern process.
Frequently Asked Questions
Build Speed Before Your Amazon OA
Timed practice is the best way to get comfortable with assessment pressure. Start with our free practice tests, then move on to coding mocks.
Keep reading
- Guide · Graduate & SchemesCorrectional Officer Exam 2026: Format, Questions & Prep
- Guide · Company GuidesAmazon Bias for Action Interview Questions & Sample Answers
- Guide · Company GuidesAmazon Insist on the Highest Standards Interview Questions & Answers
- Guide · Company GuidesAmazon Hire and Develop the Best Interview Questions & Sample Answers
- Article · Company GuidesPwC Online Assessment 2026: SHL Tests, Format & Preparation Guide
- Article · Company GuidesAmazon Assessment Test Questions & Answers (2027 Guide)