TrustBridge

Reimagining Trust. Unlocking Opportunities. A decentralized, AI-powered lending protocol bridging the gap between underserved borrowers and global liquidity.

---

🚀 What is TrustBridge?

TrustBridge is a P2P micro-lending platform that eliminates the need for traditional credit scores. By leveraging Generative AI (Gemini) for alternative data analysis and Web3 for settlement, we create verifiable on-chain reputations (TrustScores) from off-chain real-world assets (utility bills, tax returns, government IDs).

We empower individuals to own their financial identity and provide lenders with transparent, high-yield opportunities in a permissionless environment.

---

🌍 Core Ecosystem

Traditional banking excludes millions due to lack of formal credit history. TrustBridge solves the trilemma of Identity, Trust, and Access:

🤖 1. AI-Native Verification

• Document Vision: Uses Gemini 1.5 Flash to extract financial health indicators from non-standard documents (ITR, Gas/Electricity Bills, Rent Receipts).
• Biometric Liveness: Compares live camera selfies with government ID (PAN/Aadhaar) to prevent fraud using advanced computer vision.

⛓️ 2. Decentralized Settlement

• Smart Contract Escrow: Loans are requested, funded, and repaid directly on the Sepolia Testnet, ensuring immutability.
• Web3 Integration: Seamless wallet connection via Wagmi + RainbowKit creates a frictionless "Login with Ethereum" experience.

📊 3. The TrustScore Protocol

• Dynamic Reputation: A 0-100 score calculated in real-time based on verified identity (+15 pts), financial doc validity (+60 pts), and on-chain repayment history (+25 pts).
• Portable Identity: Your score lives on-chain, allowing you to carry your reputation across the DeFi ecosystem.

🔐 4. Enterprise-Grade Security

• Hybrid RBAC: Strict separation between Borrower (Request/Repay) and Lender (Invest/Track) dashboards using Firebase Auth + Custom Claims.
• Secure Storage: Sensitive PII is encrypted and stored off-chain (Cloudinary/Firestore), while only transaction hashes and scores are committed to the blockchain.

---

🛠 Tech Stack

⚙️ Frontend

• React (Vite)
• Tailwind CSS
• Wagmi + Viem (Ethereum wallet integration)

⚙️ Backend

• Flask (Python REST API)
• Firebase (Authentication + Firestore)

⚙️ AI Integration

• Gemini Vision – document parsing
• Gemini 2.5 Flash – TrustScore computation

⚙️ Blockchain

• Ethereum Sepolia Testnet
• Wagmi library (wallet connection)
• ⚠️ No smart contracts in MVP — payments are simulated

---

📊 Features

✅ TrustScore Generation

• Upload income or bill documents (PDF/images)
• AI extracts data and computes a score (0–100)
• Score improves with on-time repayments

👤 Borrower Workflow

1. Sign up and upload KYC
2. Get TrustScore
3. Apply for loan
4. Receive ETH via Sepolia wallet
5. Repay within 30 days

🧑‍💼 Lender Workflow

1. Register and post loan offers
2. Review borrower requests and TrustScore
3. Approve or reject requests
4. Access documents if borrower defaults (after 2+ months)

---

📸 Screenshots

Home Page	Dashboard	Trust Score Calculation	Profile

🔒 Privacy & Security

• 🔐 KYC and loan data stored in Firestore (not on-chain)
• 🔐 AI scoring processed server-side
• 🔐 Borrower docs shared only after default

---

🔁 User Journey

Borrower:

Register → Upload Docs → Get TrustScore → Apply for Loan → Connect Wallet → Receive ETH → Repay

Lender:

Register → Post Offer → View Requests → Review TrustScore → Approve/Reject → Track Loan

---

🧪 API Endpoints

📄 Document Parsing

• POST /vision/first-trustscore – Upload docs and generate TrustScore

📈 Trust Score Update

• POST /trustscore/update/<uid> – Update score post-repayment

💸 Loan Routes

• POST /loan/request – Request a loan
• GET /loan/user/<uid> – Fetch all user loans
• GET /loan/status/<uid>/<loan_id> – Get loan status
• POST /loan/decision/<uid>/<loan_id> – Lender approves/rejects

🏦 Lender Routes

• POST /lender/register – Register lender
• POST /lender/offer – Post a loan offer
• GET /lender/offers/<uid> – View own offers
• GET /lender/borrowers – View pending borrowers

---

🗂 Firestore Structure

users/
  └── {uid}/
        ├── loans/             # Subcollection: stores all loans requested by this borrower
        │     └── {loan_id}    # Individual loan documents with amount, purpose, status, etc.
        ├── trust_score/       # Subcollection: stores TrustScore records
        │     └── {score_id}   # Contains score value, explanation, and timestamp
        └── profile/           # Subcollection (or a document if simpler)
              └── metadata     # Contains user info like name, email, KYC flags, wallet

lenders/
  └── {lender_id}/
        ├── info/              # Subcollection: stores registration info of the lender
        │     └── metadata     # Contains PAN, interest preference, email, phone, etc.
        └── offers/            # Subcollection: list of loan offers posted
              └── {offer_id}   # Offer details like max amount, interest rate, wallet, etc.

---

🚀 Deployment

• Frontend: Vercel
• Backend: Render
• Testing: Firebase Emulator + SepoliaETH

---

🧠 Future Vision

• NFT-based identity badges
• Full loan escrow via smart contracts
• Android-first mobile app
• Real-world pilot (e.g., rural areas)

---

📮 Contact

• 📧 Email: Arnab Ghosh
• 🤝 LinkdIn: Arnab Ghosh
• 🛠 GitHub: Trust-Bridge
• 💬 Discord: Coming Soon