VisaPath

VisaPath was inspired by a simple but frustrating reality: international students often face overwhelming, fragmented, and inconsistent visa information when planning to study abroad. Many rely on scattered government websites, unofficial advice, or unclear checklists that increase stress and lead to avoidable mistakes in their applications.

The goal of VisaPath is to transform this confusion into structured guidance. The platform guides users through a step-by-step intake process where they provide key details such as destination country, nationality, visa type, funding source, accommodation details, sponsor information, and travel history.

From this input, VisaPath generates a personalized document checklist, highlights missing information, calculates an Information Completeness Score, and produces tailored visa interview preparation questions and next-step recommendations.

The system is built using a decision-support approach rather than predictive AI. It does not estimate approval chances or provide legal advice. Instead, it focuses on helping users understand requirements clearly and prepare effectively using official sources.

We built VisaPath using React, TypeScript, Vite, Tailwind CSS, and Supabase (PostgreSQL + Edge Functions). The backend logic is implemented using serverless Edge Functions that apply structured reasoning rules to user inputs.

One of the most important lessons learned was how critical Responsible AI design is in high-stakes domains. We had to carefully ensure that the system never implies visa certainty or replaces official decision-making.

Another challenge was designing a system that remains simple for users while handling complex, country-dependent visa logic behind the scenes.

Through this project, we learned how to build structured AI-driven decision support systems without relying on external large language models, and how to design transparent systems that keep humans in control of final decisions.