Generative Any Search

Home Screen
Location with Markers
Adding the Location of Food
Updation of the Quantity

Inspiration

What it does

Our aim is to try minimize food wastage by institutions, restaurants.

How we built it

We made web application using React as frontend, integrated Google Maps API, Firebase Real time database, Python server using FastAPI and CromaDB for storing vectors to improve our search queries.

Challenges we ran into

Initially, we aimed to create a one single stop for all queries irrespective of the application being targeted. We wanted to build a custom Large Language Model that could connect with any application having Rest APIs. We tried doing proof of concept using college course data but unfortuntately no college has made them public.

Accomplishments that we're proud of

We end up using vector search for our showcasing our searching ability using mathematical equations. We ended up doing a cosine similarity based implementation

What we learned

Working of Vector databases, LLAMA index, Prompt engineering, API Calls

What's next for Generative Any Search

This will be a general interface that can be integrated into any interface such as outlook and yelp and allow users not to fumble with the various options that are offered in the interface for filtering emails/ viewing top recommendations etc