Just finished the SpaceX data engineering loop. Offer declined for personal reasons but the interview itself was interesting enough to share.
Four technical rounds after an intro call. Here's what each one covered:
SQL: Two questions. One was a classic cohort retention problem (not unusual), the other was messier -- they gave you a schema that wasn't normalized well and asked you to write around it. That second one tripped me up because I kept trying to "fix" the schema in my head instead of just answering the question. Point being: work with what you're given.
Pipeline design / system design: This surprised me. It was a proper systems design round, not just "explain Airflow." They wanted to know how you'd design a data pipeline for streaming sensor data at high volume, with fault tolerance requirements and late-arriving events. I talked through Kafka, a landing zone in S3, dbt for transforms, and discussed idempotency in the processing layer. They pushed back on throughput assumptions a few times. Good engineers on the panel.
Coding: Python. Basically a LeetCode medium, data manipulation heavy. Mostly testing whether you can write clean code and reason through edge cases out loud. Not a grind-100-hards situation.
Behavioral / culture: SpaceX asks a lot of "pace" questions. They want to know you can move fast without breaking things. Come in with examples of projects where you shipped something quickly and also examples where you caught a critical issue before it hit prod. Both matter.
Comp-wise I heard SpaceX is below FAANG on base but equity can be meaningful if you believe in the trajectory. My offer was somewhere in the $160-185k base range for senior DE in LA -- not Google money but not bad for the mission.
Anyone else notice they don't really ask about data warehousing concepts? Like no Kimball vs Vault stuff. It was very applied.