DoorDash ML Interviews, Level Not Reported: 13 Experiences

Shared by janebeingjane ·
Company
DoorDash
Reports
13 interview reports
Level
Unknown
Stages
Phone screen → Onsite

DoorDash ML interview reports with no hiring level (2022–2025) mostly pair a coding question with a past-project deep dive; two asked for k-means.

At a glance

  • Process: recruiter or HR contact, then a technical screen that usually mixed coding with a project deep dive; one candidate reached a four-round onsite built around a chatbot (03).
  • Asked recently: a past-project discussion in 8 of the 10 current MLE reports and coding in 7: palindromic substrings and a chatbot design (03), an interval-value problem (04), a topological sort (08) and k-means from scratch (10, and the ML Scientist in 13).
  • Difficulty: candidate 10's k-means code wasn't runnable by the end despite hints, and 13's interviewer seemed distracted. No current report mentions an offer.
  • Practise first: All Distinct Palindromic Substrings, Maximum Coupon Credit and Its Time Windows, K-Means Delivery Zones and Detect Intent and Order Details in Help Chats.

Coverage: 13 DoorDash ML reports that never say whether the candidate was an intern, a new grad or an experienced hire: Machine Learning Engineer (12, Oct 2022 to Jul 2025, four for full-time roles, United States where stated) and Machine Learning Scientist (1, Dec 2023). Entries 11–12 (2022–Sep 2023) are more than three years old and sit under History. Other cohorts: DoorDash MLE intern interviews · DoorDash ML interviews for experienced hires.

Machine Learning Engineer

What came up

These counts cover the 10 current MLE reports (01–10).

Topic Reports Last seen
Past-project discussion or deep dive 8 of 10 Jul 2025 (report date)
Coding (data structures and algorithms) 7 of 10 May 2025 (report date)
Behavioral questions 2 of 10 Mar 2025 (report date)
Implementing k-means from scratch 1 of 10 Dec 2023 (report date)

Rows: project 01, 03, 05, 06, 07, 08, 09, 10; coding 02, 03, 04, 06, 07, 08, 10. Single reports covered a chatbot design, intent detection for a chatbot and a topological sort.

Questions we found

Question Reports Last seen
Return all palindromic substrings of a string (practice) 1: 03 Mar 2025 (report date)
Design a customer chatbot (ML system design) 1: 03 Mar 2025 (report date)
Design a chatbot from an intent classifier plus named-entity recognition (practice) 1: 03 Mar 2025 (report date)
Find the maximum value over a set of (interval, value) pairs (practice) 1: 04 Feb 2025 (report date)
A challenging project and how you overcame it (practice) (behavioral) 1: 04 Feb 2025 (report date)
For one past project: what was the problem, how did you design and choose a solution, what was the outcome, and what did you learn? 1: 05 Feb 2025 (report date)
Describe a project you've worked on 1: 06 Apr 2024 (report date)
The technical trade-offs in one of your ML projects 1: 07 Mar 2024 (report date)
A topological sort problem; the candidate no longer remembered the details 1: 08 Mar 2024 (report date)
Why this metric and not another one? 1: 09 Feb 2024 (report date)
Implement k-means clustering from scratch (practice) 1: 10 Dec 2023 (report date)

10 timelines, Dec 2023 to 2025

01: The whole screen was one project, presented to a hiring manager

Report date: Jul 31, 2025 (interview Apr–Jun 2025) · Role: Machine Learning Engineer, full-time · Background: PhD; applied online · Result: Rejected

Technical screen: project deep dive with a hiring manager

  1. The hiring manager's team works on fulfillment. The project the candidate chose to present came from ads.

The candidate called the experience positive and the difficulty average, and still didn't move forward.


02: One hour, two delivery-themed coding problems, then silence

Report date: May 13, 2025 · Role: Machine Learning Engineer · Location: United States · Result: No offer; no reply after the interview

HR contact

  1. HR got in touch to set up the interview.

Technical interview with an engineer (about 1 hour)

  1. Two medium-to-hard algorithm problems, both framed as delivery scenarios. Neither is named.

The system message said to expect a result within three weeks at most. Nothing came, and neither HR nor the interviewer answered the candidate's emails.


03: Four-round onsite built around a chatbot

Report date: Mar 20, 2025 · Role: Machine Learning Engineer

Coding

  1. Return all palindromic substrings of a given string. No constraints or examples were reported. Practice: All Distinct Palindromic Substrings.

ML system design

  1. Design a customer chatbot. No further scope was given.

ML problem solving

  1. Design a chatbot that combines an intent classifier with named-entity recognition (NER, which pulls items such as names or order details out of the text). Practice: Detect Intent and Order Details in Help Chats.

Hiring manager

  1. A basic discussion of past projects, plus behavioral questions.

04: Back-to-back technical and coding hours

Report date: Feb 19, 2025 · Role: Machine Learning Engineer · Location: United States · Result: No offer

Technical interview (1 hour, one interviewer)

  1. A challenging project and how the candidate got past it. Practice: Tell me about a challenging project that succeeded.

Coding interview (1 hour, one interviewer)

  1. A single easy problem: given pairs of (interval, value), find the maximum value. Practice: Maximum Coupon Credit and Its Time Windows.

05: A 45-minute project dive

Report date: Feb 1, 2025 · Role: Machine Learning Engineer · Location: United States · Result: No offer

Project dive (about 45 minutes)

  1. Pick one project you've already done. The interviewer asked for a lot of detail: the problem, how you designed solutions, how you chose between them, the outcome, and what you took away.

06: Emailed questionnaire, two-part technical interview, then no reply

Report date: Apr 3, 2024 · Role: Machine Learning Engineer · Location: United States · Result: No offer; no reply after the interview

Written questionnaire instead of an HR call

  1. About 15 open-ended questions to answer by email. The recruiter replied roughly two weeks after the candidate sent them back.

Technical interview, part 1: project deep dive

  1. Describe a project you've worked on. The candidate suspects this part didn't go well.

Technical interview, part 2: coding

  1. The problem isn't given. The candidate finished early, so the remaining time went into improving the solution together with the interviewer.

After the interviews the candidate followed up twice by email and never heard back.


07: LeetCode medium plus a 45-minute ML project discussion

Report date: Mar 21, 2024 · Role: ML Engineer · Location: United States · Result: No offer

Coding (60 minutes)

  1. One LeetCode question of medium difficulty; 45 of the 60 minutes were for getting the code to compile and run.

Project deep dive (45 minutes)

  1. A 45-minute discussion of one of the candidate's ML projects, including its technical trade-offs.

08: A half-remembered topological sort

Report date: Mar 1, 2024 · Role: Machine Learning Engineer · Location: United States · Result: No offer

Coding

  1. A topological sort problem in the LeetCode style; the candidate no longer remembered the details. They felt underprepared for coding.

Project deep dive

  1. Questions on past projects. The candidate enjoyed talking with the interviewers.

09: Project deep dive with a document to fill in first

Report date: Feb 1, 2024 · Role: Machine Learning Engineer · Location: United States · Result: No offer

Project deep dive

  1. Beforehand, DoorDash shared a document of questions about the candidate's background. The candidate says filling it in ahead of time helped.
  2. Questions in the round were along the lines of why a particular metric was used rather than others.

10: A k-means screen that ended without working code

Report date: Dec 8, 2023 (interview Oct–Dec 2023) · Role: Machine Learning Engineer, full-time · Background: Master's; came through a recruiter · Result: Did not pass

Rounds

  1. A phone screen in two parts: coding, then a project deep dive.

Questions

  1. Coding: write k-means by hand. Hints came, but the code still wasn't runnable by the end of the round. Practice: K-Means Delivery Zones.
  2. Project deep dive: questions on the candidate's past projects; no specific prompts are listed.

Result

  1. The candidate didn't move past the screen.

History: Machine Learning Engineer, 2022–Sep 2023

Two older reports, outside the counts above. Candidate 11 was also asked to implement k-means with fit and predict (practice).

11: k-means fit and predict in a first DoorDash screen

Report date: Sep 22, 2023 (interview Jul–Sep 2023) · Role: Machine Learning Engineer, full-time · Background: PhD; a recruiter reached out first · Result: Did not pass

This was the candidate's first interview with DoorDash, and they rated it hard.

Screen, round 1: coding

  1. Write the fit and predict functions for k-means. Practice: K-Means Delivery Zones.

Screen, round 2: hiring manager

  1. A conversation with the hiring manager. No topics were given.

The candidate thinks coding was most likely what sank it.


12: A screen that moved twice, then a no from HR

Report date: Oct 20, 2022 · Role: Machine Learning Engineer, full-time · Background: Master's; applied online · Result: Rejected

Rounds

  1. One phone screen. DoorDash first set it up as coding, then changed it to a project deep dive at a new time. Minutes before that slot the interviewer pulled out, so it was booked a third time.

Questions

  1. Project deep dive on the candidate's past work; no specific prompts are listed. The candidate thought it went well.

Result

  1. HR declined to move forward a day later with no concrete reason, which left the candidate puzzled.

Patterns across the MLE reports

  1. Coding and project talk in the same loop. Five reports pair a coding question with questions on past projects (03, 06, 07, 08, 10): a deep dive in 06, 07, 08 and 10, and a basic project discussion with the hiring manager in 03. Four reports describe only a project deep dive (01, 05, 09, 12); in 01 that deep dive was the whole screen, and in 12 it replaced a planned coding screen. A few accounts are thin, so some may describe the same candidate.

  2. k-means from scratch in 2023 screens. Two candidates had to implement k-means in the opening coding round of a two-round screen (10, 11). Report 11 names the fit and predict functions; in 10 the code wasn't runnable by the end, despite hints. Neither candidate passed the screen.

  3. Going quiet after the interview. Two candidates heard nothing back after their technical rounds despite following up (02, 06).

Machine Learning Scientist

One report (Dec 2023): a recruiter round, then k-means from scratch with an ML Scientist.

13: k-means from scratch with a distracted interviewer

Report date: Dec 30, 2023 · Role: Machine Learning Scientist · Location: United States · Result: No offer

Recruiter round

  1. A first round with a recruiter; no details given.

Technical round with an ML Scientist

  1. Implement k-means clustering from scratch. Practice: K-Means Delivery Zones.
  2. By the candidate's account, the interviewer joined 8 minutes late and never really engaged: their attention looked to be elsewhere, and several of the candidate's questions didn't seem to register.

Common questions

Does DoorDash ask ML candidates to implement k-means?

Three reports here did: two MLE screens (10, 11) and the ML Scientist round (13). Candidate 11 named fit and predict; none of the three passed.

What is the DoorDash ML project deep dive?

A talk through one past project: the problem, design choices, outcome and lessons (05), trade-offs (07) or why a metric was chosen (09). Candidate 09 filled in a document first.

Keep practicing

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.

Comments

0

No comments yet — be the first to share what you know.