Inspiration
We wanted a database of recipes that were affordable to make, shown in a dead-simple view, which tells you what to buy, where to buy it, and where to buy it for the cheapest price.
How we built it
We extracted, parsed, formatted, and ran inference on data using custom TypeScript (Deno) and C++ stacks. The main APIs we used include the OpenAI API (to extract product metadata) ScraperAPI (to pull items, pricing, and retailers)
Challenges we ran into
One of the main challenges we ran into were inconsistencies in scraped data, and actually formatting it into a usable format. Our data pipeline includes about 5 TypeScript files, with many steps that often broke and required re-parsing or re-downloading data.
What we learned
What's next for FoodFinder
Were we to continue this project, we'd likely actually implement our ideas for making the app profitable. One idea is retailer coupons, where stores can incentivize customers to come to their stores by offering special discounts (presentable at checkout), as well as affiliate revenue - if we register with large retailers and share links to products with our own tracking code, we'd make money whenever the customer makes a purchase. In both these cases, the users' and our incentives are aligned, and nobody is being lied to.
Log in or sign up for Devpost to join the conversation.