We were inspired to create an alternate system to meal kit delivery services such as Hello Fresh as while they provide convenience to consumers and somewhat reduce food waste, they involve a lot of packaging and may be out of reach for low-income families. Our system extends these ideas but incorporates the users acquiring their own produce and ingredients, accounts for eating leftovers and allows users to prioritise their health, the environment or the price of their food.
What it does
We use a proprietary algorithm and set of machine learning models to create a customised food plan for each user. This is informed by what ingredients and produce they already have, what their desired outcome is (health/env/price) and their local environment. This reduces food waste by ensuring the user can utilise their entire shopping list without food going bad. Incorporating leftovers means users can achieve their goals and help the environment without necessarily having to cook every night. Machine learning is employed to slowly learn the users cooking and purchasing habits in order to auto-fill inputs and suggest featured recipes.
How we built it
The demo is a simple flat frontend as a proof of concept but utilises real recipe data that we scraped from the internet. We used python/selenium to scrape recipe data and prices from recipe and supermarket websites and then we manually regularised and formatted this data. The frontend is a React.js webapp that uses this data to provide an onboarding and recipe planning flow for the user.
Challenges we ran into
Scraping the recipe data was challenging due to inconsistencies such as items with discounted prices, missing results for ingredients and regional spelling variation. The React app came together fairly seamlessly but we did have a few problems with caching errors in our build tool. It was also challenging to find a location to film our video pitch that had good lighting.
Accomplishments that we're proud of
We are really proud of the design and aesthetics of both our pitch and demo app. We put a lot of thought into the colour palette, logo and ux design of the app experience and the pitch. We are also proud of the engineering achievements with our web scraping scripts.
What we learned
We all learnt a lot about designing UIs, writing filming scripts and time management. Additionally, we had little experience with Selenium so we learnt a lot about that too :)
What's next for Unobtainable Magic Tesseract
Unfortunately all of us are graduating this year so we will never program again :( (hehe)