Marketplaces, gig & delivery Data Engineer Interview Loop
Data engineer interviews here are about real-time data, because dispatch, pricing and ETAs run on streaming location and order data. There's a recruiter call and a technical screen on SQL and Python. Then come four to five rounds: a SQL deep dive, coding and ETL, data modeling, system design with a focus on streaming, and behavioral. Uber adds its Bar Raiser and DoorDash adds a cross-functional interviewer. Scale counts for more than algorithms. The domain shapes every round too. Every event has a location and a timestamp, each transaction has two parties, and events arriving out of order is normal.
Companies behind this pattern
Practice by round
SQL and Python basics under time, on trip, order and driver tables. That's joins, aggregations, distinct rows and date handling, then dictionaries, lists, counting and complexity. Speed on the basics is what counts.
SQL at loop level: rides by weather and hour, the gap between booking and check-in, revenue loss by status, and the most popular room types. The pipeline questions are asked as concepts, like missing values in a dataset and filling in missing information. Expect the scale question every time.
Pipeline reasoning and projects on marketplace data. First comes handling missing and filled-in values as a design choice. Then come the projects, which are most of this round: insights from failed orders, city supply and demand, delivery market modeling and delivery duration prediction. Treat the duplicate and out-of-order records marketplaces produce as normal, because they are.
| Case study | Difficulty |
|---|---|
| Order Events: Transaction or Snapshot? | Medium |
| Trip Fact With Three Parties | Hard |
| Data project | Difficulty |
|---|---|
| Registrations by Email Provider | Easy |
| Delivery Market Modeling | Medium |
| Each User's Third Transaction | Medium |