DoorDash candidates in 2025 and 2026 reported a recruiter call, a one-hour technical screen, and a virtual onsite of about four hours with coding or Code Craft, debugging, system design and a hiring manager behavioral round. The screens were a DashMart problem with a nearest-Dasher follow-up, or a Code Craft build of an aggregator service. Onsite prompts included a refund workflow built with an AI coding tool, round-robin and consistent-hashing debugging, a notification system and a food review system.
Candidates who missed out lost time on a single bug, could not rank AI suggestions, or gave a thin project walkthrough. The one offer in these reports came from an August 2025 New York loop.
DoorDash interview process round by round
| Round | Duration | What it tests |
|---|---|---|
| Recruiter call | 10-30 min | Resume, team overlap, process |
| Technical screen | ~60 min | Code Craft build, or a coding problem with a follow-up |
| Hiring manager call (some loops) | ~20 min | Background, team fit |
| Onsite: Code Craft or coding | ~60 min | Working code; AI tool allowed in some 2026 loops |
| Onsite: Debugging | ~50-60 min | Fixing bugs in existing code |
| Onsite: System design | ~60 min | Product and logistics systems |
| Onsite: Hiring manager | 30-60 min | Projects, disagreement, mistakes, values |
DoorDash interview reports from 2025 and 2026
- Software engineer, New York (August 2025, offer). Recruiter screen, a one-hour technical screen, and four hours of onsite interviews covering system design, debugging, behavioral and algorithms, which the candidate called standard. One reported question: talk about a technical project you worked on, then look through existing code to find the bug. The Taro entry is tagged as an offer that the candidate declined.
- Software engineer (September 2025, result not stated). Vendor recruiter call, a phone screen with a senior manager, a 20-minute hiring manager call, and a virtual onsite over two afternoons. Rated hard by the candidate.
- Software engineer (August 2026, rejected). Two recruiter calls, a one-hour Code Craft screen, and four back-to-back onsite interviews. Rejected one or two days after the onsite; the candidate rated it medium difficulty.
Detailed interview experiences
Experience 1: Software engineer, New York, August 2025
Source: Taro, posted August 8, 2025. Offer received, then declined.
Recruiter screen
- A standard recruiter conversation before the technical steps.
Technical screen (1 hour)
- One live session; the candidate did not list the problem.
Onsite (about 4 hours)
- System design, debugging, behavioral and algorithms rounds.
- Reported prompt: talk about a technical project you worked on, then read through provided code and find the bug.
Takeaways
- What decided it: The candidate described every round as standard, which suggests steady performance across all four rather than one standout.
- How to practice: Rehearse a project walkthrough and a timed bug hunt in code you did not write.
Experience 2: Software engineer, two-afternoon virtual onsite, September 2025
Source: PracHub write-up published August 2026, interview in September 2025.
Vendor recruiter call (10+ minutes)
- Resume review and a check on overlap with the target team.
Phone screen (senior manager)
- A DashMart problem, solved in about 20 minutes.
- Follow-up: compute the nearest Dasher for each customer. The candidate aimed for a faster approach, settled for a slower one that loops over every customer, fixed bugs, ran the code and went slightly over time. Feedback was positive.
Hiring manager call (about 20 minutes)
- Mostly background; the hiring manager described a project.
Onsite, day 1 and day 2
- System design: Design a notification system, then handle traffic spikes.
- Hiring manager: Project experience, resolving disagreements, and a major mistake with the lesson learned. 30 minutes plus 15 for the candidate’s questions.
- Coding (Bonus Round Up): A rearrangement problem. The interviewer required the approach to be explained before any code, which took two or three attempts; the code itself took about 10 minutes. Follow-up on negative numbers.
- Project walkthrough: A senior engineer, reportedly a founding engineer, listened to a 40-minute presentation of one complex project, then 10 minutes on a second project and industry trends.
Result: Not stated. The decision arrived on the morning of the third day after the onsite.
Takeaways
- What stood out: Two separate rounds were about past projects, so the candidate’s own work carried a large share of the evaluation.
- Common fail: Coding before the interviewer accepts the approach.
- How to practice: Explain the approach in two sentences, get a nod, then code.
Experience 3: Software engineer, AI Code Craft loop, August 2026
Source: PracHub, published August 25, 2026. Rejected.
Recruiter calls
- A first call on basics and process; a second recruiter scheduled the onsite.
Technical screen: Code Craft (1 hour)
- Build an aggregator service that calls three existing services: user, payment and address.
Onsite (four back-to-back interviews)
- AI Code Craft: Implement a DAG workflow that issues refunds automatically, using an AI coding tool in a local IDE. Follow-ups on partial refunds and reducing latency. The candidate struggled to explain which of the AI’s ten suggested optimizations were practical.
- Debugging: Round-robin and consistent-hashing code. The interviewer joined about 10 minutes late, and the candidate spent most of the remaining time on a missing-letter string bug, never testing the consistent-hashing part.
- System design: Small talk, a long follow-up discussion of a past project, then a food review system with upvotes and downvotes. The candidate felt both parts went poorly.
- Hiring manager and behavioral: Questions read from the screen with little back-and-forth.
Result: Rejected one or two days after the onsite.
Takeaways
- What decided it (negatively): Time lost on one bug, and no clear ranking of AI suggestions.
- How to practice: In debugging, cap each bug at about 10 minutes. With AI tools, practice saying which suggestion you would ship and why.
DoorDash interview questions that come up often
Coding and Code Craft
- DashMart problem, then nearest Dasher for each customer
- Build an aggregator over user, payment and address services
- Refund workflow as a DAG, then partial refunds and latency
- Bonus Round Up rearrangement problem with negative numbers
- Dasher pay from base pay, tips, peak bonus and a minimum guarantee
Debugging
- Round-robin load balancer bugs
- Consistent-hashing ring bugs
- Find the bug in provided code after a project discussion
System design
- Notification system with traffic spikes
- Food review system with upvotes and downvotes
Behavioral
- Talk about a technical project you worked on
- How did you resolve a disagreement?
- A major mistake and what you learned
DoorDash interview tips from these candidates
- Get the approach approved before coding; one interviewer made the candidate restate it two or three times.
- Time-box each bug in the debugging round and test every part you were given.
- Expect your past projects to be examined in more than one round.
- If the Code Craft round allows AI tools, be ready to reject suggestions and explain why.
DoorDash behavioral round in these reports
The September 2025 hiring manager round covered project experience, disagreements and a major mistake in 30 minutes. The August 2026 round was a set of behavioral questions read from the screen. Sample answers mapped to DoorDash’s values are on the DoorDash behavioral interview questions page, and the full round list is on the DoorDash interview questions hub.
Sample answers for high-frequency questions
Q. Find the nearest Dasher for each customer.
A. On a grid, run one multi-source BFS from all Dashers at once and record which Dasher first reaches each customer cell. That is O(rows x cols) total, against O(customers x rows x cols) for one BFS per customer. Off a grid, bucket Dashers by geohash and search the customer’s bucket and its neighbors.
Q. Build an aggregator over three services.
A. Call user, payment and address services in parallel with timeouts, merge the results into one response, and decide per field what happens on failure: payment failing may fail the request, while a missing address can return a partial response with an error flag. Add retries only for idempotent reads, and write a test for each failure case.
DoorDash interview tips from these reports
Technical screen
- A September 2025 screen was a DashMart problem with a nearest-Dasher follow-up.
- An August 2026 screen was a one-hour Code Craft build over three services.
Onsite
- Debugging prompts in 2026 were round-robin and consistent hashing.
- System design prompts were a notification system and a food review system.
Behavioral
- Expect disagreement, mistake and project questions from the hiring manager.
- Some loops add a separate 40-minute project walkthrough with a senior engineer.
Sources
- Taro: software engineer, New York, August 2025
- PracHub: two-day virtual onsite after the DashMart phone screen, September 2025
- PracHub: four-round onsite with an AI-assisted coding round, August 2026
- techinterview.org: how the DoorDash engineering interview works, July 2026