GoGetScholarship

AI scholarship coach that takes students from “which scholarships?” to “ready to submit”.

GoGetScholarship helps students:

• discover relevant, realistic scholarships (not spammy lists),
• understand why each one fits them (eligibility + demographic focus),
• and get rubric-aware coaching on their essays inside a focused workspace.

Built during the Agentiiv hackathon as an end-to-end slice:
Match → Plan → Draft → Refine.

---

✨ Core features

1. Smart matching (not just a directory)

• Short onboarding collects academics, background, constraints, and goals.
• Matching engine combines:
  • Hard filters – citizenship, level of study, GPA, deadlines, etc.
  • Semantic search + LLM reranking – fit with the student’s profile and interests.
• Each scholarship card shows:
  • “Why this fits you” explanation.
  • Eligibility flags when something is uncertain or needs manual checking.
  • A rough effort estimate (essays, references, forms).

2. Planner dashboard

• Students can swipe / star / shortlist scholarships from the match page.
• Planner lays them out by deadline, with:
  • Auto-generated task checklist (essays, refs, forms, transcripts).
  • Simple workload hint so they don’t stack 5 big essays in one week.

3. Essay workspace

• Rich-text editor with an AI sidebar, not a one-click essay generator.
• AI reads:
  • The prompt and any rubric text.
  • The student profile (from onboarding).
  • Relevant context about the scholarship.
• It can:
  • Suggest outlines and story angles using the student’s real experiences.
  • Give rubric-aware feedback (clarity, impact, specificity, structure).
  • Propose line-level edits with explanations so the student learns why.

4. Winner stories (RAG-powered, WIP)

• Optional store of winner stories / example essays.
• Used as teaching material, not text to copy:
  • AI surfaces patterns (impact, quantified results, reflection).
  • Student keeps their own voice and content.

---

🧱 Tech stack

Frontend & app framework

• React + TypeScript
• TanStack Start – file-based routing, server functions
• Tailwind CSS + shadcn/ui for UI kit and design tokens

Backend & AI

• Node.js (via TanStack Start server functions)
• PostgreSQL for relational data
• Vector search (pgvector-style) for embeddings (scholarships, winner stories)
• Claude for matching explanations and essay coaching
  • Model choice is configurable via environment variables

Tooling

• pnpm (recommended) or npm / yarn
• SQL scripts for schema + seed data in sql/
• Additional helper scripts in scripts/

---

🗂 Project structure

This reflects the intended repo layout; trim folders as you clean up.

.
├── src/                # TanStack Start app (routes, components, server functions)
│   ├── routes/         # UI routes + loaders/actions
│   ├── components/     # Reusable UI components
│   ├── server/         # Server utilities (db, AI client, etc.)
│   └── lib/            # Domain helpers (matching, scoring, etc.)
├── public/             # Static assets (favicons, og images, etc.)
├── data/               # Sample data / fixtures (personas, demo profiles)
├── sql/                # SQL schema + seed scripts for Postgres
├── scripts/            # One-off scripts (ingest JSON, seed DB, etc.)
├── prompts/            # Prompt templates for matching & essay coaching
├── docs/               # Architecture, product requirements, UX flows
├── .env.example        # Template for environment variables
├── package.json
└── README.md

As you delete legacy folders, update this tree to match reality.

---

⚙️ Getting started (local dev)

1. Prerequisites

• Node.js ≥ 20
• pnpm (recommended) – or swap pnpm for npm / yarn in commands
• A running PostgreSQL instance (local, Docker, or Neon/Cloud)

2. Clone and install

git clone <your-repo-url> gogetscholarship
cd gogetscholarship
pnpm install

3. Configure environment variables

Copy the example file and fill in values:

cp .env.example .env

4. Set up the database

Run your schema + seed scripts. If you’ve wired npm scripts, this might look like:

# Example – adjust to whatever is in package.json
pnpm db:migrate    # apply sql migrations
pnpm db:seed       # seed sample scholarships + demo personas

If you don’t have CLI scripts yet, you can execute the .sql files in sql/ manually via psql or a GUI client.

5. Run the dev server

pnpm dev

Then open:

• App: http://localhost:3000 (TanStack Start dev server)

---

🧪 Demo personas

You can use these profiles when demoing the app.

Example 1 – Maya (international CS student, high need)

• 2nd-year Computer Science + Stats at the University of Toronto, GPA ≈ 3.8
• International student from Malaysia, first-generation
• Volunteers teaching Python to newcomer high-school students
• Works part-time in a campus café → wants fewer, high-value awards

Good match types:

• STEM / CS scholarships
• Women in tech / women in STEM
• First-generation and high-need awards
• Community service / leadership scholarships

Example 2 – Diego (social work, community mental health)

• 3rd-year Social Work student in Calgary, GPA ≈ 3.5
• Child of Mexican immigrants, first-gen, bilingual (English/Spanish)
• Works as a support worker + volunteers at a food bank and peer mental health program
• Needs scholarships to cut back work hours during practicum

Good match types:

• Social work / human services
• Community service and leadership
• Mental health / community impact awards
• First-generation and financial-need-based scholarships

You can store these under data/demo_profiles/ or seed them into the DB for nicer demos.

---

🏗 High-level architecture

Frontend

• TanStack Start routes for:
  • Onboarding / profile
  • Match view (swipe / shortlist)
  • Scholarship detail + “Why this fits you”
  • Planner dashboard
  • Essay workspace
  • (Optional) Winner stories browser

Backend / API

• TanStack Start server functions / API routes, e.g.:

  • GET /api/matches – fetch scholarships for a given student
  • POST /api/explain-fit – explain why a scholarship fits
  • POST /api/plan – build checklist / timeline for a scholarship
  • POST /api/coach-essay – rubric-aware feedback and suggestions

• PostgreSQL tables (example):

  • students, student_profiles
  • scholarships
  • applications, tasks
  • (Optional) winner_stories, essay_examples

• Vector columns for semantic search over scholarship text and winner-story snippets.

AI / RAG flow

1. Embeddings
   • Compute embeddings for scholarships and winner-story chunks.
2. Retrieval
   • Filter by hard constraints (citizenship, GPA, level, deadline).
   • Use vector similarity search to get candidate scholarships and examples.
3. LLM reasoning
   • Compose a prompt with:
     • Student profile
     • Scholarship fields
     • Any relevant winner-story snippets / rubrics
   • Model scores and reranks candidates, generates:
     • “Why this fits you” explanations
     • Planner suggestions (effort estimate, tasks)
     • Essay feedback and line-level edits

---

🛣 Roadmap (post-hackathon)

• ✅ MVP: end-to-end slice from onboarding → matches → planner → essay coaching.
• ⏭ Scale the scholarship dataset and clean eligibility fields for more regions.
• ⏭ Add counselor / mentor views for schools and nonprofits.
• ⏭ Track outcomes (submitted, shortlisted, won) to improve matching and prompts.
• ⏭ Explore fairness metrics for which scholarships are surfaced to which students.

---

📄 License

• MIT

---

🙋‍♀️ Contributing

This is currently a hackathon-stage codebase. Contribution are open

• The database schema.