Oldun mu?

Inspiration

We were inspired by how difficult it can be to find the right support within large communities. Especially, in universities, people often do not know who to approach or where to ask for help. Most platforms rely on broadcasting messages and hoping someone responds. We wanted to redesign that process by creating a smarter, more direct way to connect people.

What it does

It routes students to relevant communities, official resources and upcoming events based on a natural-language request. It highlights why each match fits, shows deadlines or urgency cues, and surfaces clinic or campus events when they are relevant. It also provides a community directory with filters and event maps.

How we built it

We built a Next.js web app backed by an Express API with Prisma on MongoDB. The matching pipeline combines deterministic rules with AI classification and reranking to keep results stable and explainable. On the frontend, we focused on a high-impact UI, a fast chat experience, and event mapping using MapLibre.

Challenges we ran into

Balancing relevance with safety was hard. Risk classification occurred more than we thought, so we added deterministic guards and clear explanations. We also had to keep UI performance smooth while streaming AI responses and rendering custom components, and we had to prevent SSR issues with client-only map rendering.

Accomplishments that we're proud of

A matching system that shows why each recommendation appears.
Building a stable AI matching system

What we learned

We learned that AI works best when combined with clear rules and human oversight. We also learned that community connection is often a routing problem, not a communication problem.

What's next for Oldun mu?

We want to expand the community and events catalog, add campus-specific data imports, and improve personalization using opt-in profiles. We also plan to build better feedback loops so students can rate matches and help the system learn.