Three candidates describe varied DoorDash machine learning engineer loops: two report project discussion and coding (intervals, topological sort, graphs), and the Staff candidate recalls system design and behavioral rounds.
$207K
Avg. Base Comp
$331K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
Three candidates describe DoorDash machine learning engineer loops, at ML Engineer, Senior and Staff levels. Their reports are not identical, so treat them as examples rather than a fixed format.
The ML Engineer candidate had two back-to-back one-hour technical rounds. The first was a project deep dive: about 45 minutes on a single past project, covering the problem, the design, the alternatives considered, the outcome and what was learned. The interviewer pressed on choices such as why certain metrics were picked over others. This candidate also received a document of structured background questions beforehand. The second round was a coding problem about finding a maximum value across interval and value pairs. The candidate called it easy but felt underprepared for it.
The Senior candidate reports a screen that mixed a project walkthrough with a medium problem: return a task order from a list of prerequisites, which is a topological sort. A later CodePair session ran about 45 minutes around a hard graph problem, with a case interview component also mentioned. That candidate said they did not finish the graph problem cleanly and came from a data science background.
The Staff candidate remembers the system design and behavioral rounds most, and got a generic rejection email seven days after the onsite. No question details were shared.
In practice, prepare one project you can defend in depth, including metric choices. Also practice graph traversal, dependency ordering and interval problems out loud, and rehearse system design if you are interviewing at a senior level.
Synthesized from 3 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Doordash process.
I interviewed for a Staff Machine Learning Engineer role, and the part I keep thinking about is how long the wait was. I walked out of the onsite feeling reasonably good. The system design round and the behavioral round were the two I remember most, and I'll be honest that neither was flawless. I had some small stumbles, but I don't think that's unusual at this level, since getting both of those rounds perfect is hard. Seven days after the onsite I got a generic rejection email with no feedback, which was frustrating because I'd felt the interviews went well enough for an offer. I don't have much detail on specific questions to share beyond those two round types, so take that as the extent of what I can offer. The main lesson I took away is not to read too much into how a loop feels when you're done, because a solid-feeling onsite doesn't guarantee an outcome.
Prep tip from this candidate
For a Staff-level loop, give system design and behavioral prep equal depth, since those were the two rounds this candidate singled out as the hardest to get right.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Doordash
Write a function to return True if two strings are anagrams of each other and False if they are not
| Question | |
|---|---|
| Hurdles In Data Projects | |
| Biggest Tip | |
| Unbiased Estimator | |
| Food Delivery Times | |
| How Many Friends | |
| Data Pipelines and Aggregation | |
| A Simpler KNN From Scratch | |
| Minimize Wrong Orders | |
| Ranking Metrics | |
| Client Solution Pushback | |
| Docs Metrics | |
| Food Prep Features | |
| WhatsApp Metrics | |
| Merchant Acquisition | |
| Subscription Overlap | |
| Merge Sorted Lists | |
| String Shift | |
| First to Six | |
| Bagging vs Boosting | |
| Compute Deviation | |
| 500 Cards | |
| Permutation Palindrome | |
| Raining in Seattle | |
| Encoding Categorical Features | |
| P-value to a Layman | |
| Prime to N | |
| Get Top N Frequent Words | |
| Compute Variance | |
| Weekly Aggregation |
Synthesized from candidate reports. Individual experiences may vary.
One ML Engineer candidate applied online and received a document with structured questions about their background and past work before interviewing. They found that completing it ahead of time helped frame the conversation. Other candidates did not mention this step, so it may not apply to every loop.
Two candidates report project discussion early in the process. One described about 45 minutes on a single project: the problem, design choices, alternatives, outcomes and lessons, with pushback on metric selection. The Senior candidate described a screen that combined a project walkthrough with a coding problem.
Two candidates describe coding in different forms. One had an interval and value problem in a one-hour round and felt underprepared for it. Another had a medium topological sort in a screen, then a roughly 45-minute CodePair session with a hard graph problem, and said the coding bar was where they fell short.
The Staff candidate recalls system design and behavioral rounds from the onsite, and the Senior candidate mentions a case interview component. Neither gave question details, so candidates at senior levels should plan to cover design and behavioral prep, but should not assume a particular format.
All three candidates report no offer. One followed up twice by email without a reply, and the Staff candidate received a generic rejection seven days after the onsite with no feedback. Candidates may want to plan for slow or sparse recruiter communication.
