This was the first interview of this hiring round for me, and I wasn't prepared — I underestimated how much time management the SQL portion would take, and I didn't finish it before time was cut off.
The SQL question was a real, previously-reported one. It asked for:
- Users who spent more than $100 in a month
- The same as above but excluding the person with the highest spend each month
- Month-over-month change in restaurant spend
- The ratio of order counts for restaurants in the bottom 25% by revenue
I finished the first three with ten minutes left. The interviewer pasted the query for question 4 directly and asked me to look at it and explain what it was doing. I said it was about the bottom 25% of restaurants by revenue. She then asked me to change it to calculate the ratio of order counts instead. While I was editing, she suddenly said "not the top 25%, the bottom 25%" — I was confused why she said that out of nowhere, but I realize now she'd just misremembered my earlier answer. When I went to modify the query, instead of deleting it and rewriting from scratch, I kept editing the existing query, and it got messier the more I touched it. Note to self: next time, if asked to modify a query, delete it and rewrite the whole thing yourself. Sure enough I made a mess of it, and by the time I noticed, there was only 1 minute left. The interviewer smiled and let me just explain my approach out loud instead. That said, this is a real weak spot for a DoorDash DS interview.
The case study was about bikes. This interviewer was really nice. The question itself was the same as what's in other interview reports. A few notes: the metrics need to fit the company's actual situation — don't list too many metrics, and be clear about which are primary vs. secondary. When you design the experiment, build it around the one key metric you picked, don't spread the experiment across every metric.
Overall the difficulty was moderate and the experience was decent — I was just not sharp enough that day.
Discussion
Loading comments…