Meta operates at a scale that changes standard architectural assumptions.
Users: + billion monthly active users across Facebook, Instagram, WhatsApp, and Messenger.
Data: + TB of new data ingested daily. Trillions of posts and messages stored.
Infrastructure: Custom data centers, Memcached for caching (Meta open-sourced the original), TAO (their graph database for social data), Haystack (photo storage).
Your interviewer expects you to think at this scale. Saying "we'll use a MySQL database" for a system serving billion users won't work.