DoorDash · Primly Community

DoorDash data engineer interview, pipelines and SQL: a full breakdown

de_derek · 6 replies

Just cleared the DoorDash data engineer loop (senior level, L5 equivalent) for a role on their data platform team. There's almost nothing written about what the DE loop actually looks like so I'm writing it up.

What DE interviews at DoorDash actually test (2026 loop):

SQL: Expect serious SQL. Two problems across different rounds, both involving real operational data scenarios. One involved computing dasher utilization rates across different time windows and regions. The other was a data quality problem: given a raw events table with some duplicate events and late-arriving records, write queries to deduplicate and compute a clean metric. Both required CTEs and window functions. One required CASE logic inside an aggregate. Not for the faint-hearted.

Pipeline design / architecture: This is where senior DE candidates live or die. I got a 45-minute round on pipeline architecture for the DoorDash order event stream. Questions: how do you model the ingestion layer, how do you handle late-arriving events (key for delivery data, orders update mid-flight), how do you separate raw vs transformed layers, what does your SLA look like. I talked through a medallion architecture (raw/bronze, cleaned/silver, aggregate/gold) and the interviewer pushed on how I'd handle backfills and schema evolution. Have opinions on partitioning strategies.

Coding (Python/general): One round with a Python coding problem. More practical than algorithmic: parsing and transforming a log file, handling errors gracefully. Not Leetcode-style. Think "write production-ish code quickly."

Behavioral: Same as other functions. I got asked about a time I discovered data quality issues in production and what I did about it. Have a specific story. Vague answers about "I implemented monitoring" don't land.

Stack they mentioned: Spark, Airflow, dbt, Kafka, Snowflake or Redshift depending on the team. You don't need to have used all of these but understanding the mental model of each (streaming vs batch, transformation vs orchestration) helps.

Comp offered: Around $175-180k base for senior DE, SF, 2026. Equity on top, standard four-year vest. The team moved fast: offer 8 days after onsite.

6 replies

qa_quinn

$175-180k base for senior DE, SF, 2026. That tracks with what I've seen at similar-scale marketplace companies. Equity worth pulling the full package number on before you compare to FAANG.

sre_sol

The late-arriving events question is so relevant to delivery data. An order placed at 8:58pm might have status updates rolling in until 9:45pm. If your pipeline batches hourly you have a real consistency problem. Surprised more companies don't ask this.

de_derek

Exactly. And the interviewer specifically asked how I'd handle an event that arrives 6 hours late for a metric I already published. That's where the raw layer separation pays off: you can re-derive from source. Without that you're patching published numbers which is a nightmare.

infra_ines

dbt showing up on the stack is interesting. Platform team using dbt means they probably have a lot of SQL-heavy transformation work and are trying to bring DE and analytics closer together. That's a real culture difference from teams that keep those roles siloed.

visa_vik

Did H1B sponsorship come up? Trying to figure out if DoorDash is currently sponsoring for data roles or if that's hit or miss depending on the team.

de_derek

They do sponsor but the recruiter mentioned it depends on headcount and team. Worth asking the recruiter directly in the first call. I'm a citizen so didn't dig into the details.