Meta Data Engineer interview experience

It was about an hour total with SQL and Python back-to-back, run in a plain coding environment without autocomplete. The questions weren’t individually “impossible,” but the pace was the real challenge—if I hesitated on a JOIN, I’d fall behind, and the whole session kept rolling.

The onsite was the part I hadn’t fully understood going in. I was expecting separate, clearly labeled rounds (like one pure SQL round and one modelling round), but instead it felt like the same case repeated across different product scenarios. Each interview started with product sense for about 10 minutes, then I