I don't think Uber's estimated time-to-arrival is a statistic on which a database vendor, or development team, should brag about. It's horribly imprecise.
Also isn't something that a (geo)sharded postgres DB with the appropriate indexes couldn't handle with aplomb. Number of orders to a given restaurant can't be more than a dozen a minute or so.
Especially as restaurants have a limit on their capacity to prepare food. You can't just spin up another instance of a staffed kitchen. Do these mobile-food-ordering apps include any kind of backdown on order acceptance e.g. "Joe's Diner is too busy right now, do you want to wait or try someplace else?"
The reason this happens is because Uber Eats and DoorDash and others have/had this concept where you’d “tip” for the delivery. Which is actually not a tip, but just a shitty way of disguising delivery fees and putting customers against the people that deliver the food. But that in turn has its background in how the restaurant business treats their workers in the USA, which has been wacky even long before these food delivery apps became a thing.
Anyway, regardless of your opinion on “tipping” and these practices the point was to say that there are additional complications with how much time it will take for your order to arrive aside from just the time it takes to prepare the food and the time it takes to travel from the restaurant to your door, even when the food has been prepared and a delivery driver is right there at the restaurant. If the “tip” is too low, or zero, your order could be left sitting on the shelf with nobody willing to pick it up. At least a few years ago it was like that.
All you need for this is a dictionary of zip codes and a rating -- normal, high, very high. Given that ZIPs are 5 digits, that's 100,000 records max, just keep it in memory, you don't even need entries for the "normal" ZIPs. Even if you went street-level, I doubt you'd catalog more than a few hundred thousand streets whose income is significantly more than the surrounding area.
All of this ignores the fact that adjusting a restaurant's prices by the customer's expected ability to pay often leads to killing demand among your most frequent and desirable clientele, but that's a different story.
