Design Nearby Place Recommendations

Nearby place recommendation is the location-aware discovery question: recommend somewhere a person might want to visit while their phone, and therefore their feasible choices, may be moving. Meta uses this problem to probe data collection, label quality, feature engineering, model trade-offs, and offline versus online evaluation. We follow the Interview Framework, concentrating on the parts that differ from video recommendations.

This is personalized discovery in a maps or local-discovery product. There is no search query. A nearby restaurant tab is one instance of the problem; matching typed queries to businesses belongs to search ranking.