SOTA StatWorks

From achievements to decisions in one question.

Comment

🚀 About the Project

💡 Inspiration

SOTA StatWorks was inspired by a painful reality that millions of Vietnamese students face every year:

Scholarship opportunities exist — but students don't know which ones they can actually win.

As our team observed, the scholarship application process is broken:

  • Students apply blindly to dozens of schools without knowing their real chances
  • No tool exists to predict scholarship likelihood based on individual profiles
  • Information is scattered across hundreds of university websites
  • Families waste money on "safety schools" that aren't actually safe

This creates a clear gap:
Students need data-driven guidance, not guesswork.

That's why we built SOTA StatWorks — an AI-powered scholarship prediction engine designed specifically for Vietnamese students.

"Every student deserves to know their real chances before they apply — not after they get rejected."

🧠 What We Built

SOTA StatWorks is an AI-powered scholarship prediction platform that tells students exactly which scholarships they can win — and how to improve their chances.

Instead of guessing or applying blindly, students can:

  • Upload their academic profile (transcripts, certificates, awards)
  • Get instant scholarship predictions with match levels:
    • 🎯 Dream (<40% chance) — Reach schools worth the risk
    • ✅ Target (40-65% chance) — Sweet spot for applications
    • 🛡️ Safety (>65% chance) — Guaranteed backup options
  • Understand why with AI-powered explanations
  • Run "What-If" simulations to see how improving grades affects outcomes

All in one screen, one flow, one decision.

🛠️ How We Built It

We designed the system with a clear principle:

AI handles personalization — PLS-SEM handles prediction accuracy.

Core architecture:

  • FastAPI backend for all computation
  • PLS-SEM Engine for statistical prediction (industry-standard for educational research)
  • Weighted Scoring Model based on real scholarship criteria:
    • GPA (35%), Standardized Tests (25%), IELTS (15%), Activities (15%), Awards (10%)
  • OpenAI GPT models for:
    • Student profile extraction from uploaded documents
    • Personalized recommendation generation
    • Natural language explanations
  • Web Search Integration for real-time school requirement updates
  • Next.js frontend with Vietnamese-first UX

Pipeline:

  1. Student uploads academic documents (Excel/CSV)
  2. AI extracts profile (GPA, test scores, activities, awards)
  3. Web search fetches latest scholarship requirements
  4. PLS-SEM engine computes prediction scores
  5. System categorizes schools by match level (Dream/Target/Safety)
  6. AI generates personalized improvement recommendations
  7. Simulation engine enables "what-if" scenarios

📚 What We Learned

Building SOTA StatWorks taught us several key lessons:

  • Students don't want statistics — they want clarity
    A 73.2% match score means nothing without context. We had to translate numbers into actionable "Dream/Target/Safety" categories.

  • Vietnamese education data is fragmented
    Each university has different scholarship criteria, formats, and deadlines. Building a unified prediction model required extensive research.

  • Trust is earned through transparency
    Students need to understand WHY they got a score. Our AI explanations had to be specific, not generic.

  • Real-time data matters
    Scholarship requirements change yearly. Web search integration was essential for accuracy.

  • Mobile-first is non-negotiable
    Vietnamese students live on their phones. Desktop-only was never an option.

⚔️ Challenges We Faced

1. Building a Fair Scoring Model

Different schools weight criteria differently.
We had to:

  • Research 50+ Vietnamese university scholarship programs
  • Normalize diverse criteria into a unified scoring system
  • Balance accuracy with explainability

2. Document Extraction from Vietnamese Transcripts

Vietnamese academic documents have no standard format.
Our AI had to:

  • Handle both Vietnamese and English documents
  • Extract data from tables, lists, and paragraphs
  • Validate extracted information against realistic ranges

3. Real-Time Requirement Updates

Scholarship requirements change frequently.
We integrated web search to:

  • Fetch latest GPA cutoffs and deadlines
  • Update school database dynamically
  • Flag outdated information for review

4. Latency vs Accuracy Trade-off

Students expect instant results, but:

  • PLS-SEM computation is intensive
  • Web search adds network latency
  • AI explanation generation takes time

We optimized to deliver results in under 5 seconds while maintaining accuracy.

🎯 Why It Matters

SOTA StatWorks is not just another education app.

It represents a shift:

From hoping and praying
→ to knowing and planning

By democratizing scholarship intelligence, we help:

  • Students make informed application decisions
  • Families save money on unnecessary applications
  • Universities reach qualified candidates
  • Vietnam build a more meritocratic education system

Impact potential:

  • 1M+ Vietnamese high school students yearly
  • $100M+ in scholarships available annually
  • 90% reduction in application guesswork

✨ Final Thought

"Know your chances before you apply — because every application counts."

🔗 Links

Repository: https://github.com/sotaworksvn/statworks
Live Demo: https://scholysisworks-git-edtech-etest-bernieweb3.vercel.app/
Contact: bernie.web3@gmail.com

🏆 Hackathon Information

Event: LotusHacks Vietnam 2025
Track: EdTech (Sponsored by ETEST)
Team: Phú Nhuận Builder x SOTA Works

Built With

Share this project:

Updates