Didn't see much about this online so sharing my experience. I wasn't applying for a data or BI/BA type role, so bit surprised this was part of my process. 5x questions on platform called CodeSignal. 90 minutes. First four questions worth 100 points, last one worth 200. It didn't show points per question, so budget accordingly that you'll need far more time for Q5 than the others. Its MySQL only. Revise differences if you use a different flavour -- I use Redshift every day, so some of the function differences tripped me up hard. For example, DATEDIFF in MySQL only works on dates but Redshift works on time and dates. There are half a dozen tables relating to drivers, trips, availability, etc. The tables remain consistent across all the questions, even if they aren't required, so worth spending first 10 mins reading carefully. You can run your query and get results back in a few seconds. It tells you if you are right/wrong. The test was far more about quickly getting to grips with the schema and some overly complex business request than actual SQL chops. They really should release the schema ahead of time to allow you to familiarise before tackling the questions. Anyway, i got 3/5 right (300 of 600 points). You get zero score for incomplete query, so keep that in mind if time is running short. #uberinterview #interview #uber
and this is for which role ?
Update: The questions are ranked Easy, Easy, Medium, Medium, Hard. The recruiter focussed on difficulty, asking why I only did on Medium question. Based on this, I'd suggest skipping Q1/Q2 and going straight into one of the Med and a Hard. You don't want to run out of time. I was rejected on the back of this SQL test which is strange as the role was merchant operations.
It’s not. SQL is essential for delivering good result in merchant Ops. Aside of the test execution itself which might be far from perfect I’m happy Uber is checking that. We have enough people here that “project-manage” “do strategic oversight” and other bs and are unable to do basic select * from expecting some magic “analyst team” to deliver those.
I had the same questions asked to me during a DE during a phone screen, all of the schema and tables ( half a dozen or more). All of that was shared with me “Verbally” by the interviewer and he asked 4 questions based on that all back to back, expecting me to consume all this data and 4 questions before starting to write any of my responses. The funny part is he had 2 DS/Algo questions also that he wanted to ask but never got to that since the interview was for 1 hour and I could only solve 2 and made some progress for 3rd SQL. So in essence- 6 questions ( 4 SQL, 2 DS/Algo) in 1 hour. Interviewer joins 5 min late, spends 15 mins in the introduction and explaining problems. So I have to vomit perfect code in such a short time? I felt this was very strange as no one in their right mind should be expected to write code like this and if they do they are bound to write sub-par quality code.
Ouch. Sounds shitty!