Design Uber Eats-style search function

Design the search function for a large-scale food delivery platform similar to Uber Eats.

A user opens the app and types queries such as:

• Restaurant names (e.g., "Joe's Pizza"),
• Cuisines (e.g., "Thai", "Sushi"),
• Dish names (e.g., "cheeseburger", "pad thai").

The system must return a ranked list of relevant restaurants (and possibly dishes) that can deliver to the user's location.

Requirements

Functional requirements

1. Users can search by:
   • Free-text query (restaurant name, cuisine, or dish).
   • Current location; results should be only from restaurants that deliver to that area.
2. Support filters and sort options, such as:
   • Open now / closed.
   • Price range, rating, delivery time estimate, distance.
3. Results should be ranked by:
   • Text relevance.
   • Distance / delivery ETA.
   • Restaurant popularity and rating.
   • (Optionally) user personalization.
4. Data changes that affect search:
   • New restaurants and menus being added.
   • Menu or availability changes (e.g., items out of stock, restaurant closed/open) should reflect in search reasonably quickly.

Non-functional requirements (assume a large city / multi-city deployment)

1. Low latency: P95 search latency under ~200 ms from the service (excluding network to client).
2. High availability: search should continue to work even if some components fail.
3. Scalability: support millions of users, tens/hundreds of thousands of restaurants, and high queries per second.

Describe the high-level design of such a system, covering:

• Overall architecture and major components.
• Data modeling and indexing strategy.
• Query flow from the client to returned results.
• How you keep the search index fresh when restaurants/menus/availability change.
• Approaches to ranking and personalization.
• Scaling and reliability considerations.