CraveFeed

Streamlit Docs for Image Storage using amazonS3
Streamlit docs for machine learning.
Logs of the machine recommendation model deployed on Render.
AmazonS3 Image Storage
Front page of CraveFeed
Edit profile section
Personal Profile of the user
Follow/Unfollow another user section
Main feed of the user

Inspiration

All of the teammates being foodies where in search for a website or an application that would be a one stop destination for all food lovers from all around the world so we took the matter into our own hands and created the application.

What it does

Cravefeed is built for foodies so it is primarily a social media for all food enthusiasts and connoisseurs. It allows its users to follow each other and checkout their favourite people and their tastes and preferences. Additionally, it recommends its users new food items and dishes to try based on their current dish preferences and past food choices by the hybrid machine learning model we created. Furthermore, it is especially useful to tourists and other creators who visit different places and would like a taste of the food which the locals prefer. It can also be used to connect to food bloggers and reviewers from all over the globe. Google Map integration has been done as well so you can simply navigate to the restaurant/location.

How we built it

We have used the latest tech stack for our web application. NextJS 14 + Redux for state management, tRPC for the backend, Prisma Database for the ORM, Amazon S3 Bucket for image storage, hybrid ml model which uses TF-IDF Vectorizer for our application and used Render for hosting our ML model and images.

Challenges we ran into

While we were creating the application, the documentation for tRPC and Redux integration with NextJS 14 was either non-existent or outdated. We also found several issues associated with tRPC such as the state is set before the data is resolved. We had to research about their integration from scratch. We documented all our findings in a documentation website made and hosted using Streamlit.

Accomplishments that we're proud of

We finally managed to integrate tRPC and Redux with NextJS 14. We also fixed some of the issues associated with tRPC such as the above mentioned issue. The solution to this to use set timeout inside the mutation/query call.

What we learned

We learnt new technological tools such as Redux and Amazon S3 bucket while working for this application. Attending this hackathon has helped us meet like minded individuals and tech enthusiasts.