End-to-End Multi-Step Forecasting in PyTorch

Reported in interviews at Jane Street

Problem statement

How would you forecast hourly orders for a food-delivery platform? Each night you predict the coming day using past orders, weather forecasts, holidays and planned promotions. Daily and weekly patterns are strong, with logging gaps and occasional double-counted spikes.

Choose validation and forecasting methods, then sketch a PyTorch implementation.

Show hint

Before modeling, write down exactly what is known at midnight when the forecast is made, and make every split and feature respect it.

Relevant roles: ML Engineer, Data Scientist and Applied Scientist.

Related free problems