Savora

Inspiration

We were inspired by how overwhelming choosing what to eat has become. When scrolling on Uber Eats or looking over menus, people face so many options, not able to narrow it down, and end up getting frustrated. We wanted to act on this indecision that people face choosing restaurants and what to eat at these said restaurants and turn it into something fun and intuitive. We made a visual experience where your choices form your own universe of taste.

What it does

What this solution does is visualize your food preferences into an interactive 2D star map. Each user has their own personal universe containing galaxies (restaurants), constellations (orders), and stars (individual dishes). With this, you can enter personalized prompts such as "cheap spicy Mexican food near me," and the system will reshape the map to your personal preferences. The system learns from your likes, blends in friends with similar tastes, and ultimately makes eating much simpler and more enjoyable.

How we built it

We used React to build the frontend to create a responsive and interactive 2D platform. The backend was built using Node.js and Express.js, and we used MongoDB for the database to store user and restaurant data. The Gemini API powers the AI layer, helping with prompt understanding, tag extraction, and personalized recommendations. It calculates the weightage for the prompt, user history, and social recommendations, which allow the map to be reshaped in real time.

Challenges we ran into

The biggest challenge was translating text-based preferences into a 2D visualization that felt natural. We had to balance AI complexity with a clean user experience. Getting the prompt parsing and weighting system to feel intuitive required fine-tuning the Gemini model’s outputs. We also faced challenges with data normalization and real-time rendering performance in React.

Accomplishments that we're proud of

We’re proud that we were able to take this idea that we ourselves were facing and then implement it. Through all the ups and downs, we were able to output an AI-driven 2D map that makes eating a much more enjoyable experience.

What we learned

We were able to learn how to blend AI-driven personalization with storytelling. Through this hack, we all were able to deepen our skills in prompt engineering, building an interactive frontend, and integrating large data sets with real-time visuals.