XPeng·Software Engineer·Technical Phone Screen
- In PySpark, what is the difference between row-wise and column-wise operations, and when would you use each?
- What is Parquet and why does it show up so often in analytical data pipelines?
- When would you pick MongoDB or another NoSQL solution over a lakehouse table format like Iceberg?
- When would Iceberg be the better choice over MongoDB or a traditional Hive table?
- Why would a company migrate from Hive tables to Iceberg, and what concrete improvements does Iceberg actually deliver?
“I talked through UDFs vs native column expressions and mentioned that row-wise ops are more flexible but kill parallelism if you're not careful.”