Atlassian AI coding interview: 4 candidate reports
Latest report: October 2026 · Updated
What candidates say the round looked like, summarised in our words, each linked to where it came from. Atlassian hasn’t confirmed these details unless the source says “Official page”.
Backend P40 AI-enabled coding round, about an hour in HackerRank, on an existing Java Spring Boot app with H2 and JPA and an AI assistant that was noticeably slow. The task was to add validation rules to an existing PATCH endpoint that updates a card’s PIN. The catch was the data model: the PIN was stored and accepted as an int, so leading zeros were lost, and fixing it meant changing the type through the controller, service, model and persistence layers, then proving it with tests, including that rejected updates leave the record unchanged. Their advice: know Spring Boot and JPA basics before this round.
Two steps. First, in a HackerRank link, design and explain the solution on a whiteboard, with a prompt bar to refine it using a fixed list of models (mid-size Claude and similar). Then implement it on your own laptop with any model you like; this candidate used their own Claude subscription and its top model. Others in the thread described the same flow: plan and design yourself first, then build with AI.
AI: Your own AI tools that could write code · Task: Build from a spec · Source: Reddit thread
Onsite · September 2026
Own machine
The task was to design and implement a configurable logging system from a problem statement of just 3 lines, so you have to scope the requirements yourself before designing. The candidate presented their approach, then was asked to refine it with AI and implement it for a demo. There was no given codebase: they had set up a dummy project with boilerplate before the interview, wrote an implementation document and gave it to Claude on their laptop to build.
AI: Your own AI tools that could write code · Task: Build from a spec · Source: Reddit thread
Onsite · August 2026
Passed this round
Intern · HackerRank
Summer 2027 intern offer. After an online assessment of 3 medium-to-hard questions, which the candidate found the hardest part, came an AI-assisted coding interview with the interviewer watching. A repository in the HackerRank IDE with a coding agent offering plan and agent modes and no guardrails; the task was to debug and implement some features. Difficulty was easy: the aim was to see how well you use AI to understand and fix an unfamiliar codebase. Claude Code experience makes it easy.
AI: Built-in AI chat that could read the files, write code, run tests · Task: Mixed · Source: Reddit thread