Design a personalisation layer on top of Google Search that customises result rankings based on each user's historical search and browsing behaviour, while respecting:
- •User privacy (no PII shared across requests)
- •The ability for users to opt out
- •Search quality must not degrade for cold-start users
What you'll be assessed on
How you model user intent from behaviour, privacy-preserving personalisation techniques, and the tension between personalisation and search diversity/serendipity.