Hey reader, this is my first Coderlegion article.
So I wanted a random recipe generator. Nothing fancy — point at an API, hit a button, get dinner inspiration, feel like a productive adult for once. Simple plan.
Then I actually went looking at TheMealDB (solid API, free, a lot of recipe apps run on it) and typed "Kenyan" into the search. You know what came back? Basically nothing. No ugali. No sukuma wiki. No omena. This "world food" database had somehow globe-trotted right past the food I grew up eating.
I could've just shrugged and used whatever the API gave me. I did not shrug.
The Actual Problem
TheMealDB is genuinely good for a huge chunk of global cuisine — it's just not built with East African food in mind, and I'm fairly confident I'm not the first person to notice this and move on. Most devs hit this wall and either switch APIs or just accept the gap and keep scrolling. I wanted neither. I wanted a roulette that could hand you a random pasta dish AND a legit shot at ugali night in the same spin, no hierarchy, no "the real recipes vs. the token local ones" energy.
What I Actually Built
Recipe Roulette is a hybrid random recipe generator, held together by two things:
- TheMealDB API — does the heavy lifting for the big, broad, global recipe pool. Images, ingredients, instructions, all the stuff a mature API gives you for free.
- A curated dataset I wrote myself — covers the Kenyan dishes the API completely forgot exist. Ugali, sukuma wiki, nyama choma, and more, hand-typed to match the exact same shape as the API data (ingredients, instructions, image) so when it renders, the app genuinely cannot tell which pool a dish came from. That was the whole trick — parity, not a "bonus local section" bolted on the side.
Hit the roulette button and you might get a TheMealDB classic, or you might get a dish that has never once touched a Western food database in its life. Both count. Both are equally likely. That's the point.
Why I'm Not Just Flexing a Side Project
This isn't really a "look what I built" post as much as it's a small, annoying truth I ran into: a lot of "global" APIs and datasets have blind spots you only notice once you go looking for your own food, your own language, your own context, and it's just... not there. Nobody tells you this until you build something and hit the gap yourself.
The fix didn't need to be dramatic. I didn't fork the repo or file a strongly worded issue. I just built the missing piece and merged it in at render time like it always belonged there. Because it did.
If you're building anything that touches on "global" data — recipes, place names, holidays, whatever — do the same gut check I did: does this actually include where you're from, or does it just include where the dataset's authors happened to be from?
What's Next
- Adding more underrepresented cuisines beyond Kenyan dishes (open to suggestions, seriously)
- Maybe opening it up so people can PR in their own regional dishes
- Making the mix ratio between API results and local results configurable, so it's not just 50/50 by accident
Screenshots below...
Now I Wanna Hear From YOU
Drop it in the comments:
Has your country's food ever gone missing from an API you were using?
What's the most "this dataset clearly wasn't built with me in mind" moment you've had as a dev?
Would you actually use a roulette that mixes global and local dishes, or is that just me solving my own hunger problem?
Find me on LinkedIn → Mayan Okul — come say hi.