Build model to predict package delivery time
Company: Shopify
Role: Machine Learning Engineer
Category: Machine Learning
Difficulty: medium
Interview Round: Onsite
You are building an ML model to predict package delivery time (ETA) for shipments.
Given historical shipping data (order created time, origin/destination, carrier/service level, scan events, weather, traffic, holiday seasonality, etc.), propose an approach to:
- Define the prediction target and when the prediction is made (e.g., at label creation vs after pickup vs in-transit).
- Choose model type(s) and features.
- Handle missing/late scan events and data leakage.
- Evaluate the model offline and online.
- Provide calibrated uncertainty (prediction intervals) and how you would use it in product.
- Monitor and retrain the model in production.
Quick Answer: This question evaluates a candidate's competence in end-to-end machine learning system design for time-to-event forecasting, covering prediction target definition and timing, feature engineering and model selection, handling missing or delayed signals and data leakage, uncertainty quantification, and production monitoring and retraining.
