DoorDash MLE Interview Questions: 62 Experiences (Part 1 of 2)

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

DoorDash MLE interviews since late 2023: screens shifted from coding to project deep dives, and onsites mix coding, ML design and ML knowledge.

At a glance

  • Process: recruiter call, a technical screen, then an onsite of coding, ML system design, an ML knowledge or domain round, and behavioral. In 2025–2026, 13 of the 15 screens described were a project deep dive, demo or résumé walk-through with no coding (Part 2, Patterns point 1).
  • Asked recently: a recommendation or search-ranking design in 15 of 44 reports, coding in 24, behavioral in 12. Single-Threaded CPU (LeetCode 1834) came up six times (07, 09, 17, 18, 25, 28). The newest onsite (01) added a build with an AI assistant.
  • Difficulty: design interviewers often pushed past the model into serving, storage and caching (11 reports); 15 called its scheduling problem hard with no hints. Several candidates struggled where their own work had stopped at modeling (10, 31, 32, 35, 40).
  • Practise first: Model Restaurant Recommendations, Task Order With Prerequisites, Design Personalized Restaurant Search and Build a Real-Time Prediction Service.

Coverage: 62 DoorDash experienced-hire reports in three roles: Machine Learning Engineer (60, including Senior, Ads and e-commerce teams), Senior Data Scientist, ML (1) and Software Engineer on the ML Platform team (1), Jan 2021 to Nov 2026, in New York, Seattle and unstated locations. The 44 MLE reports from late 2023 on are current: 01–23 in this part, 24–44 in Part 2. The other 18 (2021–Oct 2023) are history in Part 2, outside these counts. Other cohorts: DoorDash MLE intern interviews · DoorDash ML interviews, level not reported.

In this series

What came up

These counts cover the 44 current MLE reports in both parts. Talking through your own work and designing a recommender came up more than any single coding topic.

Topic Reports Last seen
Coding (data structures and algorithms) 24 of 44 Mar 2026 (report date)
Project deep dive, demo or résumé screen 20 of 44 Oct 2026 (report date)
ML system design 19 of 44 Nov 2026 (report date)
Recommendation or search-ranking design 15 of 44 Nov 2026 (report date)
Behavioral questions 12 of 44 Apr 2026 (report date)
Serving, deployment or production probing 11 of 44 Mar 2026 (report date)
ML knowledge questions 7 of 44 Nov 2025 (report date)
Single-Threaded CPU (LeetCode 1834) 6 of 44 Mar 2026 (report date)
ML domain case 6 of 44 Dec 2025 (report date)
Building with an AI assistant 1 of 44 Nov 2026 (report date)

Rows are built from each report's named questions and the patterns in Part 2; a report can count in several rows.

Questions we found

These tables cover all 44 current reports. Older questions are in Part 2's History.

Algorithms / Data Structures

Question Reports Last seen
Single-Threaded CPU (LeetCode 1834) 6: 07, 09, 17, 18, 25, 28 Mar 2026 (report date)
Order delivery stops or tasks so every prerequisite comes first (practice) 3: 19, 36, 39 Apr 2025 (report date)
Restaurant delivery heatmap: add each restaurant's expected deliveries to every cell in the square around it, then return the grid (practice) 2: 06, 20 Mar 2026 (report date)
Closest Binary Search Tree Value (LeetCode 270) 1: 10 Dec 2025 (report date)
An original sliding-window problem, easy to medium 1: 12 Nov 2025 (report date)
A hard problem close to Maximum Profit in Job Scheduling (LeetCode 1235) 1: 15 Jun 2025 (report date)
A simple sliding-window question 1: 23 Jan 2025 (report date)
Shortest paths that called for a specific Bellman-Ford implementation (practice) 1: 26 Sep 2024 (report date)
Return the k most popular stores, given every store's traffic 1: 27 Sep 2024 (report date)
Range sums on a 2D grid, in the style of Range Sum Query 2D (LeetCode 304) 1: 31 Aug 2024 (report date)
Largest sum over a fixed-size sliding window, framed as model scores 1: 34 Jun 2024 (report date)
A 2D version of Range Addition (LeetCode 370) 1: 37 May 2024 (report date)
Task scheduler: print when each task runs on one processor 1: 40 Apr 2024 (report date)

AI Coding

Question Reports Last seen
Build a running real-time prediction system from separate services: feature store, model registry and inference (practice) 1: 01 Nov 2026 (report date)

ML System Design

Question Reports Last seen
A recommendation system: restaurants, the home feed, a top 10 at login, grocery stores from restaurant data, or a hiring team's own product (practice) 12: 06, 07, 09, 10, 17, 18, 23, 29, 30, 31, 32, 33 Mar 2026 (report date)
Rank restaurants for a search query, or design search recommendations: features, labels and online tests; conversion versus delivery time (practice) 4: 01, 09, 18, 25 Nov 2026 (report date)
Design a food-ordering chatbot: dietary guardrails, dishes with many options, fine-tuning an open-source LLM on a budget (practice) 1: 01 Nov 2026 (report date)
Training and inference pipelines, high level and infrastructure-leaning 1: 06 Mar 2026 (report date)
Ads recommendation, where the interviewer cared about servers and data storage more than the model 1: 12 Nov 2025 (report date)
An ML problem run like a software design round (a recommender was the example given), with part of the attention on data pipelines and on storing and fetching models 1: 14 Aug 2025 (report date)
From a shelf photo and the store's catalog data, tell which items the store has in stock 1: 20 Mar 2025 (report date)
Find ads that break the platform's policies 1: 25 Oct 2024 (report date)
Match Dashers with orders, and A/B test it with time-based switchback splits 1: 33 Jun 2024 (report date)
After proposing streaming features: Kafka, database reads and writes, and QPS estimates (the design prompt isn't given) 1: 35 Jun 2024 (report date)
Explain each part concretely: how the API answers a request, the SQL query, where tables are preprocessed, and how the queue and pub/sub work (the design prompt isn't given) 1: 38 May 2024 (report date)

ML Domain Case

Question Reports Last seen
Promotions: use discount results to set prices globally, or decide which offers to send to grow the customer base (practice) 3: 10, 23, 38 Dec 2025 (report date)
An ML domain round that turned into ML system design, with probing on why the chosen models suit the problem (the prompt isn't given) 1: 14 Aug 2025 (report date)
Modeling for ads ranking 1: 17 May 2025 (report date)
Payment fraud detection, run like a data science case (practice) 1: 30 Jul 2024 (report date)

ML Knowledge

Question Reports Last seen
The bias–variance trade-off 2: 29, 30 Jul 2024 (report date)
A/B testing, which took up most of the ML knowledge round 1: 12 Nov 2025 (report date)
Convex optimization for ads bidding, down to the KKT conditions 1: 13 Aug 2025 (report date)
How do you deal with an imbalanced dataset? 1: 20 Mar 2025 (report date)
What to do once a model is in production 1: 26 Sep 2024 (report date)
About 20 quick questions built on a delivery-time prediction example (your own example is allowed), reaching into deep learning and production 1: 30 Jul 2024 (report date)
Many A/B testing questions 1: 42 Feb 2024 (report date)

Project Deep Dive

Question Reports Last seen
Have you put a model into production yourself, or served one online? 2: 32, 40 Jul 2024 (report date)
Walk through a project prepared from the recruiter's document or prompt; one interviewer asked many business-impact questions 2: 34, 36 Jun 2024 (report date)
Your role, methods, hard parts and how success was measured; trade-offs and the choice of offline, online and business metrics 1: 05 Apr 2026 (report date)
What your project did for the business after launch 1: 08 Mar 2026 (report date)
Your project's data, labels, model choice and the reason for it 1: 16 May 2025 (report date)
One project end to end: background, method, the hard parts and what you learned 1: 21 Mar 2025 (report date)
How did you pick a business threshold value? 1: 22 Feb 2025 (report date)
How and why the modeling techniques in your project work 1: 26 Sep 2024 (report date)
Why did your project use transformers, and why was deploying it hard? 1: 42 Feb 2024 (report date)

Behavioral (BQ)

Question Reports Last seen
Conflicts 4: 06, 20, 25, 26 Mar 2026 (report date)
Wrong items delivered: what could cause it, and how would you prevent it? (practice) 3: 09, 10, 30 Mar 2026 (report date)
The project you're proudest of (practice) 3: 26, 30, 31 Sep 2024 (report date)
Something you did wrong 2: 06, 30 Mar 2026 (report date)
Constructive or negative feedback you received (practice) 2: 17, 25 May 2025 (report date)
The biggest challenge you've faced 2: 19, 20 Apr 2025 (report date)
Strengths and weaknesses 2: 25, 26 Oct 2024 (report date)
What was the biggest challenge in the work you just demoed? (practice) 1: 04 Apr 2026 (report date)
How do you resolve a disagreement with a collaborator? (practice) 1: 09 Mar 2026 (report date)
What lessons did you learn from the project, and what could have gone better? (practice) 1: 08 Mar 2026 (report date)
How do you handle pressure at work or in life? 1: 17 May 2025 (report date)
A conflict with your manager 1: 19 Apr 2025 (report date)
How do you prioritize your work? 1: 19 Apr 2025 (report date)
Handling ambiguity 1: 25 Oct 2024 (report date)
Blockers you ran into 1: 26 Sep 2024 (report date)

23 Machine Learning Engineer timelines, 2025–2026

01: Onsite with an AI-assisted build, a chatbot design and search ranking

Report date: Nov 7, 2026 (interviews Oct–early Nov 2026) · Role: Machine Learning Engineer · Background: PhD, currently employed; came through a recruiting agency · Result: Rejected three days after the onsite

Onsite round 1: AI-assisted coding

  1. The exercise is a small real-time scoring system that has to run on the candidate's own laptop. A request names something like an order; one service gathers that order's inputs, another chooses the current model version, and a third returns the score. A late-delivery classifier is the example, and a stand-in model fed with made-up data is acceptable, because the grading is about the serving path rather than model quality. Every service runs separately and talks to the others through real API calls; boundaries, data shapes and edge cases are the candidate's call. Practice: Build a Real-Time Prediction Service.
  2. Expect speed over polish. The long task text went straight into the AI tool for a summary, and the session ran on quick iterations. This was a fresh task, not the delayed-order refund exercise that had been going around.

Onsite round 2: post-training evaluation

  1. Design a food-ordering chatbot for customers. Deep dives covered dietary-restriction guardrails, build-your-own dishes with many choices, and fine-tuning an open-source LLM on a small budget. Practice: Design a Food-Ordering Chatbot with Dietary Guardrails.

Onsite round 3: ML domain knowledge

  1. Restaurant search ranking, with depth on the choice of features and labels and on running the online A/B test. Practice: Design Personalized Restaurant Search.
  2. Useful background for that test: on a large change, the test and control groups share the same Dashers, which can bias the comparison. The candidate recommends DoorDash's published material on switchback experiments.

Onsite round 4: hiring manager

  1. Standard behavioral questions, each asking for a story about a real situation.

02: 105-minute phone screen: a slide-deck deep dive plus system design

Report date: Oct 20, 2026 (interview Oct 2026) · Role: Machine Learning Engineer · Background: PhD, currently employed; applied online · Result: Rejected two days later, for every level and team

Technical phone screen (105 minutes)

  1. Two parts: a deep dive into one past project, presented from the candidate's own slides, and then system design on a topic the report doesn't name.

03: Senior MLE loop with two system design rounds and a hiring freeze

Report date: Apr 15, 2026 · Role: Senior Machine Learning Engineer · Location: New York · Result: No offer

Project deep dive

  1. A high bar; the candidate says you're expected to lead the conversation.

ML system design (two rounds)

  1. Two ML system design rounds that overlapped; the prompts aren't given.

Coding

  1. A medium-to-hard coding question; the problem isn't named.

Hiring manager

  1. Behavioral questions and projects.

The schedule changed more than once, and at the end the candidate learned a hiring freeze had begun.


04: Senior AI/ML Engineer screen built around a demo

Report date: Apr 1, 2026 · Role: Senior AI/ML Engineer · Location: New York · Result: No offer

Recruiter screen

  1. A recruiter reached out on LinkedIn and held a screening call.

First round: demo of your own work

  1. The candidate presented a demo of past work and was asked what its biggest challenge had been. Closest practice: Tell me about a challenging project that succeeded.

The candidate's guess is that the interviewer wanted more recommender-systems experience than they had.


05: An hour-long project deep dive with a prep outline

Report date: Apr 20, 2026 (interviews Jan–Mar 2026) · Role: Machine Learning Engineer · Background: Master's, currently employed; referral · Result: Passed

Technical phone screen: project deep dive (60 minutes)

  1. Plan to spend the full hour on one project; there is almost no time for anything else. DoorDash sends a prep outline that asks how success was measured, what the problem was, your role and methods, and the hardest parts, so most of the preparation is a solid answer to each point.
  2. Expect questions on trade-offs and on why you picked particular offline, online and business metrics.

06: Negative-story behavioral round, a store recommender and a delivery heatmap

Report date: Mar 22, 2026 (interviews Jan–Mar 2026) · Role: Machine Learning Engineer · Background: Master's, currently employed; referral · Result: Not final when written; the candidate expected a rejection

Behavioral

  1. Expect stories about things that went badly: the examples were a conflict and a mistake you made. Having such stories ready helps.

Domain knowledge

  1. Design a system that recommends items in a store. Closest practice: Model Restaurant Recommendations.

Coding

  1. An n × n grid with restaurants on some cells. Each restaurant has a reach r and an expected delivery count d, and adds d to every cell up to r steps away, across, down or diagonally, so it covers a square that stops at the grid's edges. Return the grid of totals. Practice: Restaurant Delivery Heatmap.
  2. Working code, but it came slowly, and the candidate counted this round against them.

System design (last round)

  1. Infrastructure-leaning and high level: how training and inference pipelines are built.

07: Phone screen: a scheduling problem and a homepage recommender

Report date: Mar 20, 2026 (interviews Jan–Mar 2026) · Role: Machine Learning Engineer · Background: Bachelor's, currently employed; applied online

Technical phone screen

  1. Coding: Single-Threaded CPU (LeetCode 1834).
  2. System design: a recommendation system for the DoorDash homepage. Closest practice: Model Restaurant Recommendations.

08: One-round project deep dive focused on results after launch

Report date: Mar 12, 2026 (interviews Jan–Mar 2026) · Role: Machine Learning Engineer · Background: PhD, currently employed; referral · Result: Rejected after this round, with no reason given

Technical phone screen: project deep dive

  1. A single project filled the screen. Expect pressure on the business impact after launch and on what you learned and would change in hindsight. The session was stopped a few times for clarifications.

09: Full loop: project talk, a scheduling problem and two restaurant designs

Report date: Mar 20, 2026 (interviews Oct–Dec 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed · Result: Rejected a few weeks later

Phone screen

  1. Present a past project, with detailed follow-up questions.

Onsite round 1: coding

  1. Single-Threaded CPU (LeetCode 1834), a problem many candidates report.

Onsite round 2: ML system design

  1. A restaurant recommender where the focus was deployment rather than the model: how long-term and short-term features are kept, and which tools or stores serve them. Closest practice: Model Restaurant Recommendations.

Onsite round 3: ML discussion

  1. Rank restaurants for a search query, with depth on the model. The key trade-off: highest final conversion while keeping delivery time low. The candidate's idea, one ranker with two output heads (one per goal), did not seem to satisfy the interviewer. Practice: Design Personalized Restaurant Search.

Onsite round 4: behavioral

  1. How do you resolve a disagreement with a collaborator?
  2. What could make the food that arrives differ from what the customer ordered? Closest practice: the wrong-delivery complaints question.

10: Down-leveled after the screen, then an infrastructure-heavy onsite

Report date: Dec 19, 2025 (interviews Oct–mid-Dec 2025) · Role: Machine Learning Engineer · Background: PhD, currently employed; referral · Result: Rejected two days after the final round, without feedback

Phone screen

  1. A past project, with ML questions along the way. Expect weight on production experience: the candidate was marked down for not having deployed models online, since DoorDash expects its MLEs to handle infrastructure as well as modeling.
  2. The screen was a pass, but at a level below senior.

Onsite round 1: recommendation system design

  1. Infrastructure, not modeling: caching, scaling and the like. Closest practice: Model Restaurant Recommendations.

Onsite round 2: coding

  1. Closest Binary Search Tree Value (LeetCode 270).

Onsite round 3: hiring manager

  1. Routine behavioral questions, plus an operations question on stopping a Dasher or a store from handing over the wrong items. The expected answer was a process check at pickup, not a model. Closest practice: the wrong-delivery complaints question.

Onsite round 4: ML domain knowledge

  1. How to use promotion offers to grow the customer base. Practice: Set Discounts by Their Extra Margin.
  2. This round took place two weeks late, because the first interviewer never joined.

11: Project deep dive with an Ads team director

Report date: Dec 11, 2025 (interview Oct–Dec 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; came through a recruiting agency · Result: Rejected three days later

Technical phone screen

  1. Project deep dive: a resume walk-through led by a director from the Ads team.

12: Onsite: an A/B-heavy knowledge round and an ads recommender judged on serving

Report date: Nov 20, 2025 (onsite Oct–Nov 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; applied online · Result: Rejected the day after the onsite

Onsite

  1. Coding: an original sliding-window problem, easy to medium.
  2. ML knowledge: mostly A/B testing, closer to a data scientist screen.
  3. ML system design: ads recommendation, assessed on serving and database storage more than on the model.
  4. Hiring manager: basic behavioral questions.

13: Ads bidding screen: convex optimization down to KKT

Report date: Aug 23, 2025 (interview Jul–Aug 2025) · Role: Machine Learning Engineer, Ads team working on bidding · Background: Master's, currently employed; applied online · Result: Failed this first round

Technical phone screen (no coding)

  1. Resume-based questions, centered on convex optimization and the KKT conditions (the conditions an optimal solution must meet when a problem has constraints). The round expected depth beyond calling a solver, and that is where the candidate fell short.

14: MLE onsite with an ML design round with some focus on data pipelines

Report date: Aug 6, 2025 (interview on or before that date) · Role: Machine Learning Engineer · Background: working at Microsoft

Screen

  1. A deep dive into one of the candidate's projects.

Onsite

  1. One LeetCode-style coding round.
  2. An ML domain round, which was mostly ML system design plus questions on why they chose the models they did.
  3. ML system design run like a software design interview on an ML problem, a recommendation system for example, with attention on data pipelines. The hardest round for this candidate, who hadn't prepared for how models are stored and retrieved.

No generative AI questions came up. The candidate prepared with DoorDash's engineering blog and a system design book.


15: A hard scheduling problem with no hints

Report date: Jun 23, 2025 (interview the week before) · Role: Machine Learning Engineer · Background: PhD, currently employed; came through a recruiting agency · Result: Failed the phone screen

Technical phone screen

  1. Coding: a hard problem close to Maximum Profit in Job Scheduling (LeetCode 1235); no hints, not solved.

This came while most recent screens were project deep dives with easier coding, which is what the candidate had prepared for.


16: Project deep dive on data, labels and model choice

Report date: May 26, 2025 (interview Apr–May 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; applied online · Result: Passed

Technical phone screen (no coding)

  1. Project deep dive: the data, the labels, the model and why it was chosen, with standard ML knowledge questions in between. Here the interviewer worked in the project's own area, so the detail went deep.
  2. Questions for the interviewer: asking about DoorDash's businesses beyond food delivery landed well.

17: Senior MLE loop: a scheduling problem, an infrastructure-leaning recommender and ads ranking

Report date: May 13, 2025 (interviews Apr–May 2025) · Role: Senior Machine Learning Engineer · Background: Master's, currently employed; came through a recruiting agency · Result: Did not pass

Phone screen

  1. Project deep dive.

Onsite

  1. Coding: Single-Threaded CPU (LeetCode 1834).
  2. ML system design, infrastructure-leaning: DoorDash's recommendation system. Closest practice: Model Restaurant Recommendations.
  3. Domain knowledge: modeling for ads ranking.
  4. Hiring manager: unexpected feedback from a coworker and your response (closest practice: constructive feedback from a manager or peer), and how you handle pressure at work or in life, for example during a sharp fall in stock prices.

18: E-commerce MLE onsite with two recommendation designs

Report date: Apr 5, 2025 (interviews Jan–Mar 2025) · Role: Machine Learning Engineer, e-commerce team · Background: Master's, currently employed; came through a recruiting agency · Result: Passed

Onsite

  1. Coding: Single-Threaded CPU (LeetCode 1834).
  2. ML knowledge: standard fundamentals.
  3. ML design: restaurant recommendations for the home feed, notifications and email, judged on infrastructure rather than modeling; the interviewer managed ML infrastructure. Closest practice: Model Restaurant Recommendations.
  4. ML design: a search recommendation system. Closest practice: Design Personalized Restaurant Search.

19: Onsite where system design meant deployment

Report date: Apr 1, 2025 (interviews Jan–Mar 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; came through a recruiting agency

Onsite

  1. ML system design: detailed and about deployment and infrastructure more than the model.
  2. ML general knowledge: a knowledgeable interviewer who guided the candidate through gaps.
  3. Coding: Course Schedule-style ordering of tasks with prerequisites, about ten minutes, followed only by a complexity question. Practice: Task Order With Prerequisites.
  4. Hiring manager: a conflict with your manager, how you prioritize your work, and the biggest challenge you've faced.

20: Onsite with the delivery heatmap and a shelf-photo stock checker

Report date: Mar 24, 2025 (interviews Jan–Mar 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; applied online

Onsite (four rounds)

  1. Coding: an n × n grid with some restaurants on it. Each restaurant has a cell, a reach r and an expected number of deliveries d, and adds d to every cell within r steps in any of the eight directions, which makes a square around it cut off at the grid's edges. Return the grid of totals. The candidate's write-up came with a small 5 × 5 example with two restaurants. Practice: Restaurant Delivery Heatmap.
  2. Design: given a photo a Dasher took of store shelves, plus the store's catalog data (each product's metadata), work out which items the store has in stock and which it doesn't, much like an image search system.
  3. Domain knowledge: a detailed deep dive on earlier projects, plus basic ML questions such as how to deal with an imbalanced dataset.
  4. Hiring manager chat: past projects and the candidate's role, then behavioral questions including conflict and a challenge you faced.

21: A 45-to-60-minute deep dive with a familiar story arc

Report date: Mar 12, 2025 (interview Jan–Mar 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; applied online · Result: Passed

Technical phone screen

  1. Project deep dive, 45 to 60 minutes on one project after introductions. Cover the background, the problem, the method, what got hard and how you handled it, and what you took away, in any order; expect technical detail questions along the way. The interviewer helped structure the story when it wandered.

22: A deep dive walked through from a prepared outline

Report date: Feb 28, 2025 (interview Jan–Feb 2025) · Role: Machine Learning Engineer · Background: Master's, currently employed; came through a recruiting agency · Result: Passed

Technical phone screen

  1. Project deep dive on one recent project, after brief introductions. The candidate's outline (background, training samples, features, model, serving, evaluation) set the order of the discussion.
  2. Moderate follow-ups, for example how a business threshold was set; the answer was a data-science estimate first, tuned in later versions.

23: Causal ML for pricing: strong coding and design, then a missed domain round

Report date: Jan 9, 2025 (interview on or before that date) · Role: Machine Learning Engineer, scope: causal ML for pricing · Background: L5 Applied Scientist at Amazon, no PhD · Result: Failed the domain round of the virtual onsite

Virtual onsite: coding

  1. A simple sliding-window question. The candidate felt good about it.

Virtual onsite: ML system design

  1. A design about recommendations. The candidate proposed a two-tower network, with embeddings kept in a vector database for fast retrieval, and felt good about this round too. Closest practice: Model Restaurant Recommendations.

Virtual onsite: ML domain knowledge

  1. The case: if DoorDash offers a promotion such as x% off or $y off, how can it use what it learns to set prices globally? Practice: Set Discounts by Their Extra Margin.
  2. The candidate proposed regressions that estimate the effect of the promotion while controlling for customer clusters and product clusters. The interviewer kept nudging them toward a much broader answer, but the candidate never worked out what that meant. This is the round they failed.

The candidate later wondered if the role really meant reinforcement learning for pricing.


Common questions

Does the DoorDash MLE phone screen have coding?

Not usually any more. In 2025–2026, 13 of 15 described screens were a project deep dive, demo or résumé talk; only 07 and 15 had coding. Older screens were mostly coding (Part 2).

What system design does DoorDash ask MLE candidates?

Mostly recommendation or search ranking: restaurants, the home feed, store items or search results (15 of 44). In eight of those, the interviewer cared most about serving, storage or caching (09, 10, 12, 17, 18, 29, 30, 31).

Which coding problem does DoorDash repeat for MLE?

Single-Threaded CPU (LeetCode 1834), in six reports from mid-2024 to early 2026 (07, 09, 17, 18, 25, 28), at the screen or the onsite. Task ordering with prerequisites came up three times (19, 36, 39).

How should I prepare the DoorDash project deep dive?

Several recruiters sent a document or prompt to prepare from (05, 26, 34, 36), and 22 wrote their own outline. Interviewers asked about business impact after launch (08), metrics (05) and whether you put models into production yourself (32, 40).

What to practise

Keep practicing

In this series

Sources: Anywhere useful interview info can be found.

Next in this seriesDoorDash MLE Earlier Interviews: 39 Experiences (Part 2 of 2)
0 total reactions. Your reactions: none.

Comments

0

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