finished the Atlassian data engineer interview loop last month, reporting in.
full disclosure: i didn't get the offer. but i got detailed feedback which was surprisingly useful, so i'll share both the process and what they told me afterward.
loop structure for DE (senior level, ~7 YOE): recruiter screen SQL and data modeling round pipeline design round behavioral round cross-functional round (this was added late, caught me slightly off guard)
SQL and data modeling (60 min): heavier on modeling than raw SQL. they gave me a scenario: design a data model for tracking user activity across Confluence and Jira. I had to normalize it, identify key entity relationships, think about slowly changing dimensions for user account state. the SQL questions layered on top: given this model, how would you write a query to get weekly active users per product with 30-day trailing window.
the interviewer was calibrating on how i think about grain (row-per-event vs. pre-aggregated fact tables) and whether i could talk about trade-offs between normalization and query performance.
pipeline design round (60 min): design a pipeline to ingest webhooks from Jira, process and enrich the events, land them in a data warehouse for analytics. this is basically a streaming ingestion problem. they cared about: idempotency, schema evolution, handling late-arriving events, monitoring. i went with Kafka to BigQuery via a consumer service. the idempotency discussion was the deepest part.
where i got dinged: their feedback was that i was too opinionated about tooling ("i would use Kafka here") without first establishing the constraints. they wanted me to ask what the event volume was, whether we needed real-time or near-real-time, what the existing infrastructure was. i jumped to solutions. fair critique, applies broadly to pipeline design interviews.
overall: Decent interview experience. well-organized, feedback was real, not generic. the bar for DE felt higher than average mid-market companies, roughly comparable to Databricks or Snowflake from what i've seen.