← DoorDash Interview Insights

DoorDash·Data Scientist·Technical Phone Screen·Senior

SeniorPrefer not to say
Apr 2026Remote

Summary

DoorDash data science interview that went deep on ML system design for delivery prediction. The whole thing was basically one long case study and it covered way more ground than I expected, from feature engineering to deployment guardrails.

Questions Asked (6)

Q1

How would you define the target variable for a late delivery prediction model, and how would you structure a time-based train/validation/test split to avoid leakage?

Product Analytics & MetricsData ModelingTechnical Trade-offs
Author's notes

The target definition sounds obvious until you realize there are edge cases.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the target variable clearly: a binary label indicating whether a delivery exceeded its estimated delivery time, with careful handling of edge cases like cancellations. Then explain how to structure a time-based split that respects the temporal order of data, using a rolling or expanding window approach to prevent leakage from future data into training.

Pro tip: Mention that you would use the delivery's original estimated time (not a post-hoc adjusted one) and consider excluding deliveries with missing or unreliable timestamps to avoid label noise. Also, emphasize that the split should mimic the production scenario where you predict future deliveries using only past data.

1. Define the target variable

Specify a binary label: 1 if actual delivery time > estimated delivery time (late), else 0. Clarify how to handle cancellations, missing data, and whether to use a threshold (e.g., >5 minutes late) based on business impact.

2. Choose the time-based split strategy

Use a chronological split: train on the earliest data, validate on a subsequent period, and test on the most recent period. Alternatively, use a rolling window approach for multiple splits to assess model stability over time.

3. Ensure no leakage

Verify that all features are computed using only data available at prediction time (e.g., no future delivery times, no aggregated stats that include the validation/test period). Also, avoid using the target variable in any feature engineering.

4. Validate the split

Check that the distribution of key features and the target variable is similar across splits, and that the time periods are contiguous without gaps. Consider using a gap between train and validation to simulate real-world delay in label availability.

5. Iterate and monitor

After deployment, continuously monitor model performance over time and retrain with new data, using the same time-based split logic to avoid leakage in future evaluations.

Key Points to Mention

  • Binary classification target: late vs. on-time, with clear definition of 'late' based on estimated delivery time.
  • Time-based split: train on past, validate on future, test on most recent, to mimic production.
  • Leakage prevention: ensure features are computed only from past data, no target leakage.
  • Rolling window or expanding window approach for robust evaluation.
  • Handling of cancellations and missing data: exclude or treat as separate class.
  • Business context: align target definition with DoorDash's delivery promise and customer expectations.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

What are at least 10 production-safe features you would engineer at order-creation time for this late delivery model?

Data ModelingProduct Analytics & MetricsTechnical Trade-offs
Author's notes

This is where I actually felt decent.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the late delivery model and its business context, then structure your answer around key areas like data quality, model robustness, and operational safety. For each feature, explain how it prevents production issues such as data leakage, bias, or system failures, and tie it back to DoorDash's logistics challenges.

Pro tip: Emphasize features that enable monitoring and graceful degradation, as these are often overlooked but critical for production systems. Also, quantify the impact where possible to show business acumen.

1. Clarify the model and context

Ask clarifying questions about the late delivery model, such as its definition, data sources, and how it integrates with order creation. This ensures your features are relevant and shows you think before coding.

2. Categorize features by safety aspect

Organize your features into categories like data validation, model robustness, real-time performance, and monitoring. This makes your answer structured and comprehensive.

3. Detail each feature with rationale

For each feature, briefly explain what it does and why it's production-safe, focusing on preventing failures or ensuring reliability. Use examples from similar systems if possible.

4. Highlight trade-offs and monitoring

Discuss any trade-offs (e.g., latency vs. accuracy) and how you would monitor the features in production to detect issues early. This demonstrates maturity in deploying models.

Key Points to Mention

  • Data validation and schema checks to prevent garbage inputs
  • Feature freshness and time-travel correctness to avoid leakage
  • Fallback logic for missing or low-confidence predictions
  • Real-time performance constraints and latency budgets
  • Monitoring and alerting for model drift and data quality
  • A/B testing and shadow deployment for safe rollout

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Identify at least five leakage hazards in this dataset and explain how you would eliminate each one.

Data ModelingTechnical Trade-offsRoot Cause Analysis
Author's notes

Leakage questions are my kryptonite because they seem easy and then you miss the subtle ones.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the dataset's context and the modeling goal, then systematically identify leakage hazards by examining features for temporal, target, or data collection dependencies. For each hazard, propose a concrete elimination strategy, such as removing the feature, adjusting the data split, or using proper cross-validation.

Pro tip: Emphasize that leakage often arises from subtle data collection artifacts, so always validate your fixes by checking if model performance drops to a realistic level—if it doesn't, you might have missed a leak.

1. Understand the Data and Goal

Ask clarifying questions about how the data was collected, what the target variable is, and the prediction time frame. This sets the stage for identifying leaks.

2. Identify Potential Leakage Hazards

List at least five hazards, such as target leakage, temporal leakage, train-test contamination, leakage from data preprocessing, and leakage from group structures.

3. Explain Elimination Strategies

For each hazard, describe a specific method to eliminate it, like removing features, using time-based splits, fitting preprocessing only on training data, or using group-aware cross-validation.

4. Validate the Fixes

Discuss how you would verify that leakage is eliminated, such as by monitoring performance changes or conducting ablation studies.

Key Points to Mention

  • Target leakage: features that include information about the target that wouldn't be available at prediction time.
  • Temporal leakage: using future data to predict past events; fix with time-based splits.
  • Train-test contamination: preprocessing steps like scaling or imputation fitted on the entire dataset; fix by fitting only on training data.
  • Group leakage: when data from the same entity appears in both train and test; fix with group-aware splitting.
  • Data leakage from feature engineering: creating features that inadvertently use future information; fix by carefully designing features using only past data.
  • Validation strategies: use of pipelines, cross-validation with time series splits, and holdout sets to detect and prevent leakage.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

Which evaluation metrics would you use for both ranking and calibration, and how would you build a cost matrix to justify operational thresholds?

Product Analytics & MetricsA/B Testing & ExperimentationTechnical Trade-offs
Author's notes

AUROC and AUPRC for ranking, Brier score and calibration slope for calibration, that part came out fine.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by distinguishing ranking metrics (e.g., NDCG, MAP, MRR) from calibration metrics (e.g., Brier score, log loss, calibration curves), then explain how to build a cost matrix that weighs false positives and false negatives based on business costs. Finally, show how to use the cost matrix to derive an optimal threshold that minimizes expected cost, and discuss how this applies to DoorDash's delivery time or search ranking problems.

Pro tip: Tie the metrics and cost matrix directly to DoorDash's business objectives—e.g., optimizing delivery time predictions to reduce late deliveries (false negatives) vs. overestimating time (false positives) that hurts customer experience. Mention that thresholds should be dynamic and monitored via A/B tests.

1. Define ranking and calibration metrics

For ranking, use NDCG, MAP, or MRR to evaluate order of items; for calibration, use Brier score, log loss, or reliability diagrams to assess probability accuracy.

2. Identify business costs of errors

Quantify costs of false positives (e.g., overestimating delivery time leading to customer dissatisfaction) and false negatives (e.g., underestimating time causing late deliveries and refunds).

3. Build a cost matrix

Create a 2x2 matrix with costs for true positives, true negatives, false positives, and false negatives, using dollar values or business impact scores.

4. Derive optimal threshold

Use the cost matrix to compute the expected cost at different thresholds and select the threshold that minimizes total expected cost.

5. Validate and iterate

Test the threshold in A/B experiments, monitor both ranking and calibration metrics, and adjust as business costs change.

Key Points to Mention

  • NDCG and MAP for ranking; Brier score and calibration curves for calibration.
  • Cost matrix should incorporate asymmetric costs (e.g., late delivery costs more than early).
  • Threshold optimization via expected cost minimization.
  • Use of A/B testing to validate threshold changes.
  • Dynamic thresholds based on context (e.g., peak hours, restaurant type).
  • Trade-off between ranking quality and calibration accuracy.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q5

Walk through how you would ramp this model to production: shadow mode, treatment gating, A/B testing with guardrails, drift monitoring, and recalibration strategy.

A/B Testing & ExperimentationSystem DesignTechnical Trade-offs
Author's notes

Shadow mode to log predictions without acting on them, then gradual rollout with a hold-out control group, guardrails on key metrics like refund rate and dasher utilization.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Structure your answer as a phased rollout plan, starting with shadow mode to validate model behavior without user impact, then treatment gating to control exposure, followed by A/B testing with guardrails to measure causal impact, and finally drift monitoring and recalibration to sustain performance. Emphasize how each phase de-risks the next and ties back to business metrics like delivery time or conversion.

Pro tip: At DoorDash, always tie your rollout plan to marketplace dynamics—e.g., how model changes affect Dasher supply, customer wait times, and merchant operations—and mention that you'd monitor guardrails like order cancellation rate and Dasher utilization, not just model accuracy.

1. Shadow Mode

Deploy the model to log predictions in real-time without affecting user experience, then compare its outputs against the current production model or business outcomes to catch obvious errors and validate infrastructure.

2. Treatment Gating

Gradually expose a small, random subset of traffic to the new model (e.g., 1-5%) while holding back the rest, using feature flags to control rollout and monitor system health and early business metrics.

3. A/B Testing with Guardrails

Run a statistically powered A/B test with pre-defined primary metrics (e.g., delivery time) and guardrail metrics (e.g., cancellation rate, Dasher utilization) to measure causal impact and ensure no harm.

4. Drift Monitoring

Continuously track input feature distributions, prediction distributions, and model performance metrics in production, setting alerts for significant deviations that could indicate data drift or concept drift.

5. Recalibration Strategy

Define triggers and cadence for model recalibration (e.g., when drift exceeds threshold or performance degrades), and decide whether to retrain, fine-tune, or adjust thresholds, with a rollback plan if issues persist.

Key Points to Mention

  • Shadow mode for safe validation and infrastructure testing without user impact
  • Treatment gating with feature flags and gradual rollout to limit blast radius
  • A/B testing with guardrail metrics (e.g., cancellation rate, Dasher utilization) to prevent negative marketplace effects
  • Drift monitoring for both data drift (input features) and concept drift (model performance)
  • Recalibration triggers and cadence, including automated retraining pipelines and rollback procedures
  • Business metric alignment (e.g., delivery time, order volume) and statistical power considerations

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q6

How would you handle cold-start problems for new restaurants or new cities, and how would you account for seasonality in the model?

Data ModelingTechnical Trade-offsAdaptability & Ambiguity
Author's notes

Hierarchical pooling was my answer for cold start, basically borrowing signal from similar stores or zones when you have no history.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing the cold-start problem as a hierarchical modeling challenge where you borrow strength from similar restaurants or cities, then progressively personalize as data accumulates. For seasonality, explain how you would incorporate time-based features and use techniques like seasonal decomposition or Fourier terms to capture recurring patterns. Emphasize the importance of validating with holdout data and monitoring model performance over time.

Pro tip: Mention that cold-start and seasonality often interact—new restaurants in new cities may have no historical seasonality data, so you might need to rely on global seasonality patterns initially and adjust as local data arrives. Also, highlight the business impact: accurate cold-start predictions can improve user experience and restaurant partner success from day one.

1. Define the problem and success metrics

Clarify what we're predicting (e.g., demand, delivery time) and how we'll measure success (e.g., RMSE, business KPIs). This ensures alignment with stakeholders.

2. Leverage hierarchical/global models for cold-start

Use a hierarchical Bayesian model or a global model with restaurant/city embeddings to share information across entities. For completely new entities, fall back to segment-level or global averages.

3. Incorporate seasonality explicitly

Add time-based features such as day-of-week, month, holiday indicators, and Fourier terms to capture periodic patterns. Consider seasonal decomposition or STL for time series models.

4. Design for progressive personalization

As data accumulates for the new restaurant or city, dynamically update the model to give more weight to entity-specific data, reducing reliance on global patterns.

5. Validate and monitor

Use time-based cross-validation to evaluate performance, especially for cold-start scenarios. Set up monitoring to detect drift and retrain as needed.

Key Points to Mention

  • Hierarchical models or mixed-effects models to borrow strength across groups
  • Feature engineering for seasonality: Fourier terms, holiday flags, interaction with location
  • Use of embeddings for restaurants and cities in neural networks or matrix factorization
  • Fallback strategies: global averages, similar restaurant/city matching
  • Time-based cross-validation to avoid leakage and assess seasonality handling
  • Business considerations: impact on delivery estimates, restaurant onboarding, and user trust

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.