Amazon SWE Interview: Online Assessment Guide

Estimated read time: 7-9 minutes

Summary: The Amazon SWE online assessment is one of the clearest parts of the public Amazon SDE process, especially for university candidates and SDE II. For SDE II, the research describes a 90-minute coding section, about 15 minutes of system design scenarios, about 10 minutes of work style questions, and an optional survey. This guide explains how to treat the OA as a multi-signal screen rather than just a coding test.

TL;DR + FAQ (read this first)

At-a-glance takeaways

• For SDE II, the official research describes 90 minutes of coding, about 15 minutes of system design, about 10 minutes of work style questions, and an optional survey.
• University and new-grad SDE paths also use an OA as an early step.
• The OA is strongest evidence for New Grad/SDE I and SDE II. SDE III+ should treat it as possible or role-dependent.
• The coding section is not the only signal. Amazon is also checking systems thinking and Leadership Principle-style work habits.
• Do not prepare only for syntax. Prepare for scalable code, edge cases, maintainability, and consistent work-style judgment.

Quick FAQ

Is the Amazon SDE OA required for everyone?
The research is strongest for university/new-grad SDE and SDE II. It says SDE II candidates must complete the OA. SDE III+ exact use is less clear.

Is the OA only coding?
No. For SDE II, the research includes coding, system design scenarios, work style, and an optional survey.

Who evaluates it?
The OA itself is automated or asynchronous, then successful candidates move toward the interview loop.

What is the biggest mistake?
Treating the OA as two isolated coding questions and ignoring maintainability, scalability, and work-style consistency.

---

1) What the Amazon SDE OA checks

The Amazon research describes the SDE II OA as a gate before interviews. It collects several signals at once: coding ability, basic system-design judgment, work style, and role matching.

The coding section is reported as 90 minutes with two questions. The system design portion is shorter, about 15 minutes, so expect scenario judgment rather than a full whiteboard design. The work-style section is short too, but it matters because Amazon's loop is built around Leadership Principles.

Takeaway: the OA is a compressed version of Amazon's broader hiring bar. It asks whether you can code, reason about systems, and show work habits that fit the loop.

---

2) Questions and tasks you may face

Exact OA questions are confidential, so the research gives official task types and topic areas rather than a fixed question bank. Phrase your practice around the tasks Amazon is actually collecting signal on.

• Complete two coding questions in 90 minutes, then explain the time and space complexity of each solution.
• Given an array or string problem, write logical, maintainable code and handle edge cases before submitting.
• Given a graph or data-structure problem, choose an approach that scales beyond the simplest input size.
• Given a short system design scenario, choose the most robust breakdown and explain the tradeoff.
• Given an object-oriented design task, model the core entities and operations clearly.
• Given a database or distributed-computing scenario, identify the likely bottleneck and the safer design choice.
• Given a work-style situation, choose the response that best reflects ownership, customer impact, and sound judgment.
• Given an ambiguous implementation task, state assumptions, solve the core case, and account for failure paths.

---

3) Level-specific expectations

Relevant levels: Intern, SDE I/New Grad (L4), SDE II (L5), SDE III+ where role-dependent.

For intern and new-grad candidates, the OA is a central early technical filter. For SDE II, the source is especially specific: coding, system design, work style, and survey components are described. For SDE III and above, the public evidence is weaker, so treat the OA as possible rather than guaranteed.

If you are targeting SDE II, do not skip system design preparation just because the OA is online. The system-design portion is short, but it still signals whether you can reason beyond code.

---

4) Evaluation signals

Strong OA performance shows syntactically correct code, scalable problem breakdown, robust edge-case handling, and maintainability. In the system design scenarios, strong candidates identify the core tradeoff quickly instead of overbuilding. In the work-style section, consistency matters.

Weak performance often looks like code that only handles the happy path, solutions that do not scale, or work-style choices that conflict with ownership and customer focus. Because the OA is asynchronous, your submission has to speak for itself.

Do this now: practice explaining the simplest correct solution, then the scalable version, then the tradeoff between them.

---

5) Common failure modes

Ignoring the system design component. The SDE II OA includes about 15 minutes of system design scenarios. Treat it as real signal.

Optimizing only for passing sample tests. Amazon's research emphasizes logical, maintainable, scalable code. Passing the obvious cases is not enough.

Rushing work-style answers. Work-style questions are short, but they connect to Leadership Principles that recur in the loop.

Assuming all levels get the same OA. University and SDE II evidence is strong. SDE III+ is less certain.

Not managing time across two coding questions. A polished first answer does not rescue an empty second answer.

---

6) How to prepare

• Practice two-question timed coding sessions, not single untimed problems.
• Review arrays, strings, graphs, object-oriented design, databases, distributed systems basics, operating systems, and internet fundamentals.
• For each solution, write the complexity and one edge case before moving on.
• Practice short system-design scenarios where you choose between two reasonable tradeoffs.
• Review Leadership Principle-style work situations, especially ownership, customer impact, and earning trust.

The best OA prep is balanced. You need speed, but also enough judgment that the code and work-style choices look like Amazon loop evidence.

---