Scholarly

Live deployment: https://hack-the-scholarship-frontend.vercel.app/ 🌟 Inspiration Every year, billions of dollars in scholarships go unclaimed - not because students are unqualified, but because the application process is confusing, exhausting, and overwhelmingly generic. Students are told: “Write a compelling essay.” But they’re never told what makes an essay compelling for a specific scholarship. A merit scholarship wants academic rigor. A service scholarship wants community impact. A leadership scholarship wants initiative. Yet students write one essay and send it to everything. We asked ourselves: “What if AI could read a scholarship the way a human judge does? What if it could uncover what the scholarship truly values - and then help students tell the right story for the right audience?” That insight became the foundation of Scholarly - the AI-powered system that helps students not just find scholarships, but actually win them. 🚀 What we built Scholarly is a fully deployed , end-to-end AI scholarship application platform that helps students: Create a reusable, structured personal profile Analyze any scholarship automatically Generate tailored application essays based on scholarship-specific priorities Understand why the Claude AI wrote each section (explainability matrix) Add new scholarships simply by pasting a URL - our scraper handles the rest We also built a chrome extension that autofills the applications for you too! 🔍 How it works (Our Pipeline) We designed a multi-stage AI pipeline using LLMs: Scholarship Personality Analysis LLM extracts hidden values, patterns, and success traits. Adaptive Weight Generation Each scholarship gets a custom weight profile Student Profile Structuring The system organizes the student's background into a structured vector: Targeted Essay Drafting Using scholarship weights × student strengths, Claude AI generates a perfectly tailored essay. Explainability Matrix Every sentence is tied to a scholarship value and a user strength. Scholarship Scraper Integration Paste any URL → scrape → clean text → run analysis → generate essay. 🛠️ How we built it We built ScholarAI as a full-stack platform: Next.js + TypeScript for the front-end TailwindCSS + shadcn/ui for a modern, clean UI PostgreSQL + Prisma for persistent user profiles & applications Node API routes for LLM orchestration Claude 3.5 Sonnet for the multi-stage AI reasoning pipeline Recharts for adaptive weighting visualizations Custom Web Scraper for extracting scholarship text Vercel Deployment for production-level hosting 🔮 What's next for Scholarly LLMs are incredible at pattern recognition, but only when given structure. Explainability matters. Students trust AI more when they understand why it writes what it writes. Modularity beats mega-prompts when building reliable AI systems. UX is everything - especially for stressed students applying to scholarships.