While brainstorming, we tried to come up with personal pain points that we experienced on a daily basis and that would also be felt by many other individuals other than just us. After coming up with a few ideas, we settled on a software solution that allows users to have better control over their grocery shopping and cooking experience.

What it does

The app is composed of three main screens: The first screen on which the user lands is a digital representation of the food stocked in his kitchen. Broken down between two components, Pantry and Refrigerator, the interface stores all the food items the user owns. The second screen offers the user the option to scan a grocery store recipe by either taking a picture of it or uploading it from the camera roll. Our software then identifies the items on the list and creates both a record of the items purchased and updates the food stock accessible on the starting screen. Finally, the third screen allows the user to generate custom recipes based on the items available in their stock using a third-party AI model.

How we built it

  • Used React for the frontend
  • Used Python for the backend and image processing
  • Used PostgreSQL to store the food stock
  • Amazon Textract model used to scan and read receipts
  • ML models from Cohere and OpenAI to categories items and generate recipes

Challenges we ran into

  • Time Crunch. The 24-hour time limit made it challenging to implement all the features we wanted to
  • Integrating all the features we created together
  • It was hard to focus on the project with all the fun workshops going on :)
  • Learning how to implement Receipt Scanning

Accomplishments that we're proud of

  • Learned a lot about ML technologies and NLP
  • Got a working product
  • Made new friends
  • Gained experience in React and Flask

What we learned

  • Learn from each other since we all had different skillsets
  • How to use OpenAI chat as an API
  • Fixing Flask CORS errors

What's next for Chelf!

As for the next steps, we would focus on fine-tuning our main features and implementing additional nice-to-have features to create a seamless end-to-end experience.

Share this project: