The 1st round was an online assessment with 30 BigQuery questions to be solved in 40 minutes. I was able to solve 26 questions due to time constraints. The overall question levels were easy-medium difficulty.

In the 2nd round, the interviewer couldn’t explain the question well and couldn’t tell what he expected from me. When I asked him what he typically looks for in a candidate, he replied, “logic.” He also mentioned he doesn’t care about coding; he looks for logic only and asked about FARM_FINGERPRINT in BQ to create a Surrogate Key Hash. It seemed hypocritical to say he doesn’t care about coding and ask a question on a niche hash function. The interview revolved around data modeling concepts and scenarios, like strategy to implement Surrogate Key, Star vs. Snowflake, Fact vs. Dimension, and implementing Views, Slowly Changing Dimensions II.

The 3rd round revolved around my projects and day-to-day work and the current architecture of the data platform I work on. Questions were asked around query optimization and disaster recovery on the production environment, and Data Lake and Lake house design in my team.

The 4th round they called me to reappear for the online assessment because, during the first round, my face was not visible properly. This time, I was able to solve 30 questions with 30 seconds remaining.

SCD Types, Query Optimization, Data Modelling, Data Lake, Handling Production Environment Crashes, and Trade-offs.