Voluntree
A campus volunteering platform that fills event slots fairly.
- Year
- 2026
- Role
- Team lead and developer · team of four
joinvoluntree.vercel.app
The problem
Volunteer slots at AIU go to whoever replies first in WhatsApp and Telegram groups. In a 20-student survey, 7 of 13 applicants had been turned away at least once, and 4 of 13 volunteers never received a certificate.
The solution
Voluntree replaces the race with a transparent process. It recommends students who have had fewer opportunities, lets the database enforce capacity, and closes the loop with verified attendance and a certificate. It was built for the MMC Upstart 2026 Catalyst (Focus Area 02, Smart Campus and Institutional Systems).
- AI drafts a mission from a pasted event announcement; a person always edits and publishes it.
- A transparent score recommends students with fewer past opportunities and explains why ("Why recommended").
- Capacity is enforced by the database, so there is no race for slots.
- QR check-in and organizer-approved attendance, then a PDF certificate and growth of the student's private date palm.
- Admin view with a fairness chart, a dean-approval queue and logged revocations.
Key screens
Landing page: fair volunteering at AIU. 
Each leaf on the tree is a mission a student showed up for. 
AI drafts the mission from an announcement; the organizer checks every field before publishing. 
Fair matching, with reasons: every recommendation says why.
Mobile.
Technical decisions
Overbooking is impossible at the database level
PostgreSQL row-level security and locking functions mean capacity is never exceeded, and a repeated approval changes nothing. The schema is built from 12 Supabase migrations and backed by 293 automated tests.
AI drafts, people decide
A Groq LLM step turns an announcement into a mission draft. Its output is validated with Zod, and a human must review it before anything is published.
Results
Across a simulation of 100 students and 20 missions, Voluntree placed 88.9% of first-time volunteers, against 66.7% under first-come-first-served. It was fairer on 10 of 10 further random seeds.
The trade-off: the most experienced 10% of students receive fewer new placements.
Voluntree won the 2nd Place Prize in Catalyst: Igniting Future Innovation (MMC Upstart 2026).
Note: The evaluation uses synthetic data, not real students.
Stack & credits
- Next.js (App Router)
- TypeScript
- Tailwind CSS
- Supabase (PostgreSQL, RLS)
- Groq LLM
- Zod
- pdf-lib
- Vitest
- Vercel
Built by a team of four. The matching, certificate and evaluation modules were built by my co-developer.