Atlassian SWE: Online Assessment or Karat Prescreen Guide

Estimated read time: 7-9 minutes

Summary: Atlassian SWE candidates may see an online assessment, a Karat-style prescreen, or another early technical filter depending on level, team, and pipeline. The research is strongest for early-career online coding and structured reports of OA or Karat-style screens. This guide explains how to prepare for a timed, early technical gate without assuming it replaces later live coding.

TL;DR + FAQ (read this first)

At-a-glance takeaways

• This stage is possible for interns, new grads, junior, mid-level, and senior+ candidates, but it is role-dependent.
• Official early-career evidence supports an online coding test and several virtual interviews for graduate paths.
• Structured candidate reports mention OA or Karat-style screens before deeper technical rounds.
• Expect data structures, practical coding, communication, edge cases, and time management.
• Senior candidates should confirm whether this stage is actually in their loop, because design and leadership rounds may carry more weight later.

Quick FAQ

Is the OA guaranteed?
No. Treat it as possible, especially for early-career paths, then follow your recruiter packet.

Is Karat the same as an OA?
No. Karat-style screens are live or semi-live technical screens, while online assessments are usually timed and self-directed.

What should I optimize for?
Readable correctness first, then edge cases, complexity, and calm pacing.

---

1) How the early technical screen works

The research groups this stage as an online assessment or Karat prescreen because Atlassian paths vary. For early-career candidates, expect a timed online coding test. For other pipelines, candidate reports describe a technical prescreen before the final loop.

The important distinction is control. In an OA, you manage time alone. In a Karat-style screen, you may need to explain your reasoning live while coding. Both formats reward the same habits: clarify the task, write simple code, test edge cases, and avoid overbuilding.

---

2) Question styles you may face

Exact Atlassian-owned question wording is not reliably public. These examples are representative of the data-structures and practical-coding signal described in the research.

• Given a stream of events with user IDs and timestamps, return the users who exceeded a rate limit in a rolling window.
• Given a list of tickets with dependencies, return a valid order to complete them or report that a cycle exists.
• Implement a small in-memory key-value store with get, put, delete, and optional expiry.
• Given a grid of project tasks, find the shortest path that satisfies dependency and blocked-cell constraints.
• Given a list of document edits, merge non-overlapping ranges and detect conflicting updates.
• Implement a simple autocomplete or prefix lookup, then discuss how memory usage changes as the dataset grows.

---

3) What strong signal looks like

Strong candidates avoid panic-driven complexity. They state assumptions, solve the core case, handle edge cases, and explain why the approach fits the constraints. If the screen is live, they keep the interviewer oriented while coding.

For early-career candidates, fundamentals matter most: arrays, maps, strings, graphs, recursion, sorting, and basic complexity. For senior candidates who receive this screen, the bar may include cleaner structure and more confident tradeoff discussion even if the problem itself is not senior-level design.

---

4) Common failure modes

Chasing the clever solution first. Start with a correct baseline before optimizing.

Not testing boundary cases. Empty input, duplicates, cycles, expired data, and conflicting updates are common places to fail.

Letting the platform drive your pacing. In timed settings, reserve time for tests and cleanup.

Treating the prescreen as the whole loop. Passing this gate does not remove later coding, design, manager, or values signal.

---

5) How to prepare

• Run 45-60 minute timed sessions with one normal case and two edge cases written before coding.
• Practice maps, heaps, queues, graphs, sliding windows, and simple object modeling.
• For live screens, narrate assumptions and complexity while you work.
• For OA-style screens, keep code small and testable rather than clever.
• Ask your recruiter whether this is online-only, live, Karat-style, or skipped for your role.