Fit Selector

• 

  website demo

Inspiration

The project started with a problem that our teammate Edward had from time to time: you can't see everything in your wardrobe at the same time, and if you're the type of person that likes having a lot of options, it's likely that you've bought similar/near duplicate stuff before. If you could keep track of everything in your wardrobe by simply uploading and using natural language to search through it, you could not only save yourself time and money, but improve your ability to throw fits together!

What it does

It takes clothes that the user uploads, indexes it using AI, and all you need to do to search through your wardrobe is a simple prompt that brings up the top X most relevant pieces of clothing

How we built it

We used FastAPI to connect all the AI and semantic search DB stuff together to the frontend, and used standard React + Vite for UI stuff. Gemini and MongoDB provided the AI semantic search capabilities, and Clerk saved us time for authentication.

Challenges we ran into

Integration hell really came through this time.

Accomplishments that we're proud of

We spent less than a day on authentication, which is a monumental achievement for something aiming to be an MVP under 36 hours. We also learned MongoDB and the Gemini API rather quickly.

What we learned

Trying to go for a whole spectrum of tracks and challenges will dilute your efforts and the scope of the project will outrun actual progress very quickly. Solidifying the scope from the get go, and sticking to it is crucial.

What's next for Fit Selector

The AI prompt is currently a work in progress, but once finished will allow an AI to put together your next outfit, given the weather, occasion, or anything else you'd like to tell it. It would also walk you through its thinking process about why it chose the pieces it did.

Built With

• gemini
• mongodb
• python
• react