Skip to content

marlinprotocol/AI-Job-Matching-dApp

BranchesTags

Repository files navigation

Decentralized Job Marketplace with AI-Powered Matching & Enclave Verification

A comprehensive blockchain-based job marketplace that implements both traditional O(n²) matching and advanced AI-powered matching using Llama 3.2 with Marlin Oyster CVM for secure, verifiable AI computations. The platform connects job seekers with employers through intelligent analysis and cryptographic verification.

🚀 Key Features

Core Functionality

  • Smart Contract-Based: Decentralized job postings and candidate profiles on Ethereum
  • Dual Matching Algorithms: Traditional rule-based O(n²) and AI-powered intelligent matching
  • Wallet Integration: Seamless MetaMask connectivity for secure transactions
  • Modern UI: Next.js TypeScript frontend with responsive Tailwind CSS design

AI-Powered Matching System

  • Llama 3.2 Integration: Advanced language model for intelligent candidate-job analysis
  • Marlin Oyster CVM: Secure, verifiable AI computations with cryptographic proofs
  • Enclave Verification: On-chain verification of AI responses using smart contracts
  • Detailed Analysis: Comprehensive reasoning and candidate evaluation beyond simple scoring

Advanced Features

  • Job Detail Pages: Individual job viewing with complete information
  • Skill Extraction: AI-powered skill extraction from job descriptions and resumes
  • Real-time Verification: Live verification of AI responses with signature validation
  • Attestation System: Cryptographic attestation for AI computation integrity
  • Header Extraction: Oyster signature and timestamp capture for verification

🏗️ Project Architecture

AI-Job-Matching-DApp/
├── Contracts/
│   ├── JobMarketplace.sol          # Main contract with O(n²) matching
│   └── LlamaResponseVerify.sol     # Enclave verification contract
├── app/
│   ├── api/
│   │   ├── ai-candidates/          # AI matching endpoint
│   │   ├── attestation/            # Attestation API
│   │   ├── generate/               # LLM generation API
│   │   └── oyster/                 # Direct Oyster API proxy
│   ├── components/
│   │   └── Header.tsx              # Navigation component
│   ├── deploy/                     # Verification testing page
│   ├── jobs/
│   │   ├── page.tsx               # Job listing page
│   │   └── [id]/page.tsx          # Individual job details
│   ├── matches/                    # Dual algorithm matching page
│   ├── post-job/                   # Job posting interface
│   ├── register-seeker/            # Job seeker registration
│   ├── setup/                      # Contract setup page
│   └── utils/
│       ├── contractHelper.ts       # Contract interaction utilities
│       ├── llmService.ts          # AI and verification services
│       └── web3.ts                # Ethereum integration
└── public/                        # Static assets

Getting Started

Prerequisites

  • Node.js (v16+)
  • npm or yarn
  • MetaMask extension installed in your browser

Frontend Setup

  1. Navigate to the frontend directory:
cd AI-job-Matching-dApp/
  1. Install dependencies:
npm install
  1. Run the development server:
npm run dev

  1. Open your browser and navigate to http://localhost:3000

  2. On first use, go to app/utils/web3.ts, Enter the deployed contract addresses, create .env file according to the example and connect your MetaMask wallet.

🎯 Using the Application

As an Employer

  1. Connect Wallet: Connect your Ethereum wallet (MetaMask)
  2. Post Jobs: Navigate to "Post a Job" and fill in details
    • Job title, description, and requirements
    • Required skills (with AI-powered skill extraction)
    • Location and salary information
  3. Browse Applications: View candidate matches through the AI matching system
  4. Track Verification: See cryptographic proofs of AI analysis integrity

As a Job Seeker

  1. Connect Wallet: Connect your Ethereum wallet (MetaMask)
  2. Create Profile: Navigate to "Register as Job Seeker"
    • Personal information and skills
    • AI-powered skill extraction from resume text
    • Location and salary expectations
  3. Browse Jobs: View all available positions with detailed job pages
  4. Get Matched: Benefit from both traditional and AI-powered matching

Job Browsing & Details

  1. Browse All Jobs: Navigate to "Browse Jobs" to see all active positions
  2. View Details: Click "View Details" on any job for comprehensive information
  3. Smart Filtering: Jobs are automatically filtered to show only active positions
  4. Real-time Updates: Job information is fetched directly from the blockchain

Advanced Matching & Verification

  1. Navigate to Matches: Go to "View Matches" page
  2. Choose Algorithm:
    • Traditional O(n²): Fast, rule-based matching with on-chain storage
    • AI-Powered: Intelligent analysis with Llama 3.2 and cryptographic verification
  3. AI Matching Process:
    • System calls Llama 3.2 model via Marlin Oyster CVM
    • Captures Oyster signatures and timestamps
    • Performs enclave attestation for public key verification
    • Verifies AI responses on-chain using smart contracts
  4. View Results:
    • Traditional matches show scores and basic reasoning
    • AI matches include detailed evaluation and verification status
  5. Verification Status: See cryptographic proof of AI computation integrity

Testing Enclave Verification

  1. Deploy Page: Navigate to "Deploy on Oyster" for verification testing
  2. Test Known Values: Uses hardcoded working values to verify contract functionality
  3. Debug Tools: Comprehensive logging for troubleshooting verification issues

🔍 Matching Algorithms

Traditional O(n²) Algorithm

  • Compares each job with each candidate systematically
  • Calculates scores based on:
    • Skills matching (50% of total score) - exact string matching
    • Location matching (30% of score) - exact location match
    • Salary matching (20% of score) - salary range compatibility
  • Fast and deterministic results
  • Stored on-chain in smart contract

AI-Powered Matching (Enhanced)

  • Uses Llama 3.2 LLM running on Marlin Oyster CVM
  • Intelligently analyzes job descriptions and candidate profiles
  • Considers context, synonyms, and semantic relationships
  • Provides detailed reasoning for each match
  • Offers comprehensive candidate evaluations on demand
  • Scores candidates holistically beyond simple rule matching
  • NEW: Cryptographic verification of AI responses
  • NEW: On-chain verification using enclave signatures

🔐 Enclave Verification System

How It Works

  1. AI Computation: Llama 3.2 processes job-candidate matching in secure enclave
  2. Signature Capture: Oyster headers (signature + timestamp) are captured from AI responses
  3. Attestation: System performs cryptographic attestation to get enclave public key
  4. Receipt Creation: ABI-encoded receipt data combines prompt, response, and context
  5. On-Chain Verification: Smart contract verifies the signature using the enclave public key

Verification Components

  • Oyster Headers: x-oyster-signature and x-oyster-timestamp from AI responses
  • Receipt Data: ABI-encoded data containing model, prompt, context, response, and tokenization
  • Enclave Public Key: Retrieved through attestation process
  • Smart Contract: LlamaResponseVerify.sol contract performs cryptographic verification

🛠️ Development

Smart Contract Architecture

JobMarketplace.sol

  • Main Functions: Job posting, candidate registration, O(n²) matching
  • Matching Algorithm: calculateMatches() function implements traditional scoring
  • Score Calculation: Skills (50%), location (30%), salary (20%)
  • Data Storage: On-chain storage of jobs, candidates, and match results

LlamaResponseVerify.sol (EnclaveResponseVerifier)

  • Verification Function: verifyEnclaveResponse() validates AI computation signatures
  • Cryptographic Verification: Uses ECDSA signature verification
  • Receipt Validation: Validates ABI-encoded receipt data format
  • Integration: Works with Marlin Oyster CVM verification system

Frontend Architecture

  • Framework: Next.js 15 with App Router and TypeScript
  • Styling: Tailwind CSS for responsive design
  • Blockchain: ethers.js for Ethereum integration
  • State Management: React hooks for component state
  • API Routes: Server-side API endpoints for AI and verification services

LLM Integration and AI Matching

Setup Requirements

  1. Copy env.example to .env.local
  2. Set NEXT_PUBLIC_LLAMA_INSTANCE_IP to your Llama instance IP
  3. Ensure Marlin Oyster CVM is properly configured
  4. Deploy both JobMarketplace and EnclaveResponseVerifier contracts

Available AI Features

  1. Skill Extraction: Extract skills from job descriptions and resumes
  2. AI-Powered Matching: Intelligent candidate-job matching with reasoning
  3. Detailed Evaluation: Comprehensive analysis beyond simple scoring
  4. Cryptographic Verification: On-chain verification of AI computations
  5. Oyster Integration: Secure and verifiable AI computations

Example API Calls

Basic LLM Query:

curl http://{{instance-ip}}:5000/api/generate -d '{
  "model": "llama3.2",
  "prompt":"What is Ethereum?"
}'

AI Candidate Matching:

// Via frontend API with verification
const result = await fetch('/api/ai-candidates', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({ jobs: jobsArray, candidates: candidatesArray })
});

// Result includes matches and verification data
const { matches, verificationData } = await result.json();

Enclave Verification:

// Verify AI response on-chain
const verified = await verifyEnclaveResponse(
  receiptData,    // ABI-encoded receipt
  timestamp,      // Oyster timestamp
  signature,      // Oyster signature
  enclavePublicKey // From attestation
);

🚀 Deployment

Contract Deployment

  1. Deploy JobMarketplace.sol to your target network
  2. Deploy LlamaResponseVerify.sol for enclave verification
  3. Update contract addresses in web3.ts
  4. Set up Marlin Oyster CVM instance

Frontend Deployment

  1. Configure environment variables
  2. Build the Next.js application: npm run build
  3. Deploy to your preferred hosting platform
  4. Ensure MetaMask connectivity for users

🔧 Configuration

Environment Variables

NEXT_PUBLIC_LLAMA_INSTANCE_IP=your-llama-instance-ip
NEXT_PUBLIC_CONTRACT_ADDRESS=your-contract-address
NEXT_PUBLIC_VERIFIER_CONTRACT=your-verifier-address

Network Configuration

  • Ensure MetaMask is connected to the correct network
  • Contract addresses must match deployed instances
  • RPC endpoints should be properly configured

🏆 Key Innovations

  1. Dual Algorithm Approach: Traditional + AI matching for comprehensive coverage
  2. Cryptographic Verification: On-chain verification of AI computations
  3. Marlin Oyster Integration: Secure enclave computations with proof generation
  4. Comprehensive Analysis: Beyond simple scoring to detailed candidate evaluation
  5. Real-time Verification: Live verification of AI responses with signature validation
  6. Modular Architecture: Separate contracts for core functionality and verification

About

This project was developed as a Hands-On demo project for the Basics section of "Offchain computing using TEE coprocessors" course.

Resources

Readme
Activity
Custom properties

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages