Your are indeed correct, it is indeed an BMR calculation, specifically I use a revised Harris-Benedict equation.
https://en.wikipedia.org/wiki/Harris–Benedict_equation#Calcu...
Perhaps though I should by default use a TDEE calculation...
For weight loss, eating at BMR is about right generally. And I'd imagine most use cases for calorie tracking apps are about weight loss.
To make it most accurate, I'd personally just add a basic multiplier for what sort of lifestyle people have (sedentary, small exercise like walking the dog, moderate exercise and hard exercise in the multipliers of 1.2x, 1.4x, 1.6x, 1.8x).
I exercise a lot and that typically adds 1000-1500 calories onto my daily BMR of 1500 or so, so the multipliers a lot of BMR / TDEE calculator websites use seem to make sense.
I'm more curious where the food database is sourced from. I tried something similar by getting data from British supermarkets (Sainsbury's, Tescos etc.) and none of them seemed to have APIs, so web scraping was the only possibility.
As for the food db: I use Open Food Facts which is great, but can be messy and sometimes wrong as it's mostly crowdsourced. I combine that data with data from US Food Data Central, which is an awesome resource.
You will see open food facts data as unverified in the app whilst food data central will appear as verified. Hope that makes sense.
I am in the process of adding major US restaurant chains as I recently found another great dataset with more than 30000 entries. So McDonalds, Starbucks, Dunkin Donuts, etc also coming.