CommonGrounds

Homepage
Coffee Mode
Matcha Mode
Profile
Create Event

CommonGrounds is a swipe-based platform that helps students explore opportunities that reflect who they are and who they're becoming!

Identity in Action

Identity is built on interaction. And in a world that tends to funnel us into the same mundane cycles (like grinding and networking for internships), we tend to forget that there are so many other experiences out there. CoffeeGrounds aims to help students break out of the mundane.

Personalize Your Blend

CommonGrounds allows users to post and discover both personal experiences and professional projects; others can like or skip based on their interests. The platform is split into two distinct modes.

🍵 Matcha focuses on hobbies, passions, and growing your personal identity.
☕Coffee focuses on career interests and professional growth through peer-driven projects.

AI Personalization

As users interact with the app, CommonGrounds uses an AI-driven personalization algorithm to learn from behavior. By analyzing user profiles, event descriptions, swiping patterns, and time spent, the system adapts recommendations to better align with each user's evolving identity over time.

Reflective Dashboard

The platform also features a reflective dashboard that provides statistics for users to analyze for themselves. Additionally, the dashboard contains an AI insight that suggests potential new experiences to pursue or a new ways to strike that perfect balance.

Technical Aspects

The database is managed through Supabase. The recommendation algorithm is built with two different models/agents:

Gemini 2.5 Flash-Lite, which analyzes current user statistics and provides meaningful outputs for both the recommendation system and the user's insights
all-MiniLM-L6-v2, a sentence-transformers model that converts user data and event data into useful embeddings, which, using cosine similarity, allows for a quick and efficient recommendation system Embeddings for users are constantly updated using the data collected and the insights from Gemini. This acts as a feedback loop, where recommendations impact user decisions, and user decisions impact recommendations. The frontend was created using Next.js and was connected to the backend using FastAPI.

What's Brewing for CommonGrounds

In the future, we hope to track more data in order to utilize stronger analysis tools like the tools Amplitude provides, which will allow for better recommendations and greater user satisfaction. Additionally, we hope to build location functionality, as well as a potential chat option.