Atlassian makes tools that teams use to plan and track their work, such as Jira for work items, Confluence for shared pages, Trello for boards and Bitbucket for code. As a fresher, prepare to explain basic data structures through these products: a card moving across a board, an issue blocking another issue, a search across a project. Bring stories from college group work. What to show here: clear thinking about how teams plan, track and share work. Each question below has what the interviewer is checking, an answer frame and a sample spoken answer to adapt.
How Atlassian hires, in its own words, and its questions for every job →
“In my final year project, four of us built your project. In the second week two of us wrote the same login screen, and nobody had started the database. We were tracking work in a chat group, so tasks got buried. I made a simple board with three columns, to do, doing and done, and wrote one card per task with one owner. I asked everyone to move their own card before our evening call, so the call became short. After that, your result. I learned that a list only helps if updating it takes a few seconds.”
Weak answers blame lazy teammates and show no change in how the group tracked its work.
“During my internship, setting up the project on a new laptop took most of a day because the steps lived in people's heads. When I finally got it running, I wrote every step on a shared page, including the errors I hit and how I fixed them. I kept it short, with commands that could be copied. When the next intern joined, I asked her to follow the page without my help and I fixed the parts where she got stuck. The team then linked it in the readme. The result was your result.”
Weak answers describe a long document nobody read, or never tested the guide on a real reader.
“I would treat it as a small state machine. I would keep a map where each state points to the set of states it can move to. For example, In progress points to In review and To do. One function takes the current state and the wanted state and looks in the map. If the move is not there, it returns a clear error. Because the rules are data, a team that wants an extra state can change the map without me rewriting the logic. In my college project I used an enum and a map for something similar.”
Weak answers write a long chain of if statements for every pair of states and cannot say how a new state would be added.
“The simple way is to give each card a position like one, two, three. But if I drop a card in the middle, every card below must be updated. A better way is to leave large gaps between positions, so a new card takes a value halfway between its two neighbours, and only that one card is saved. After many moves the gap between two cards can run out. Then I would respace that column once in the background. I could also use a linked list, but finding the order for display would need more work.”
Weak answers renumber the whole column on every drag and do not notice the cost on a long column.
“I would see issues as nodes in a graph and each blocks link as an arrow. When a user adds a link from one issue to another, I would start at the second issue and follow its outgoing arrows using depth first search, keeping a set of visited issues. If I ever reach the first issue, the new link would close a loop, so I reject it with a message that names the chain. The visited set also stops me from walking the same issue twice. The cost grows with the number of linked issues, which is usually small.”
Weak answers check only the direct pair of issues and miss loops that pass through several issues.
“Scanning every title means the time grows with the number of issues, and it repeats for every search. Instead I would build a map ahead of time, where each word points to a list of issue ids that contain it. A search for one word is then a single lookup. For two words, I take both lists and keep the ids that appear in both. When someone edits a title, I remove the old words for that issue and add the new ones. I would also store words in lower case so that case does not matter.”
Weak answers only say use a database and cannot explain what an index does underneath.
“I would not wait a full day, and I would not build on a guess. First I would read the linked tickets, past comments and the area of code it points to. Then I would write on the ticket what I understand, in two or three lines, and list my questions so each can be answered shortly. While the reporter is offline, I would do the work that is needed whatever the answer is, such as setting up the code locally and writing a test that shows the current behaviour. I would also ask a teammate nearby if they know the background.”
Weak answers either wait silently for a reply or build the whole feature on an unchecked guess.
“In every group project I have done, the hard part was rarely the code. It was knowing who was doing what and where the decisions were written. I have used Trello for a college project and Jira in my internship, so I have seen how a simple board can calm a team down. Building tools like that means my users are teams like the ones I have been in, and I understand their pain. I also like that the problems mix data structures with how people behave. I want to learn how to build software that many teams depend on daily.”
Weak answers talk about office perks or fame and cannot name what the products do.
No figures on this page: pay depends on the level, the city and the offer in front of you. When the offer comes, the free salary negotiation script helps with that part.
ClapAssist is an AI interview assistant for Mac and Windows. It listens to the interview on your computer and shows you what to say, in short lines you can read while you talk. Your live interview audio and screen are never stored. Your resume and notes are saved to your account so the app fills them in on any computer. It is hidden from screen share on every plan, including Free; only you can see it.