AI Code Evaluation Game for Training Purposes

URL

https://github.com/copocaneta/code-evaluation-platform/

Overview

The AI Code Evaluation Game is an interactive platform that combines coding challenges with real-time AI evaluation using GPT-4. This educational tool helps developers improve their coding skills through immediate, intelligent feedback and features an extensible challenge system for customized learning paths.

Key Features

🤖 AI-Powered Code Evaluation

• Real-time code assessment using GPT-4
• Dual-phase evaluation system:
  • Comprehensive code analysis
  • Binary pass/fail determination
• Intelligent feedback without revealing complete solutions

🎯 Learning Features

• Multiple difficulty levels for progressive learning
• Instant feedback for continuous improvement
• Points-based reward system
• Real-time leaderboard tracking

💻 Technical Implementation

// Example of the evaluation result structure
interface EvaluationResult {
  id: string;          // Unique identifier
  timestamp: string;   // Evaluation timestamp
  status: 'success' | 'error' | 'warning';
  content: string;     // AI feedback
}

Extensible Challenge System

Challenge Structure

interface Challenge {
  id: string;           // Unique identifier for the challenge
  title: string;        // Display title
  description: string;  // Challenge requirements
  defaultLanguage: string;  // Initial programming language
  defaultCode: string;  // Solution example/template
  initialCode: string;  // Starting code for users
}

Adding Custom Challenges

Challenges can be easily added by modifying the challengeLoader.ts file:

const challenges: Challenge[] = [
  {
    id: "custom-challenge",
    title: "Your Challenge Title",
    description: "Description of what needs to be accomplished",
    defaultLanguage: "python",
    defaultCode: 'def solution():\n    # Example solution\n    pass',
    initialCode: `# Starting point for users\ndef solution():\n    pass`
  }
];

Training Scenario Examples

1. Code Refactoring Training

{
  id: "reducing-verbosity",
  title: "Reducing Verbosity",
  description: "Replace excessive use of loops featuring multiple temporary variables with functional patterns.",
  defaultLanguage: "python",
  defaultCode: '# Example of clean, functional code...',
  initialCode: `# Code that needs refactoring...`
}

2. Design Pattern Implementation

{
  id: "singleton-pattern",
  title: "Implement Singleton Pattern",
  description: "Create a thread-safe singleton class with lazy initialization.",
  defaultLanguage: "python",
  defaultCode: '# Example of correct singleton implementation...',
  initialCode: `# Basic class structure to be modified...`
}

3. Algorithm Optimization

{
  id: "optimize-search",
  title: "Search Optimization",
  description: "Optimize the search algorithm to achieve O(log n) complexity.",
  defaultLanguage: "python",
  defaultCode: '# Optimized binary search implementation...',
  initialCode: `# Linear search implementation to be optimized...`
}

Training Applications

1. Corporate Training

   • Custom challenges for company-specific coding standards
   • Team-specific best practices
   • Internal tool usage training

2. Academic Settings

   • Data structures and algorithms exercises
   • Programming language syntax practice
   • Software design principles

3. Interview Preparation

   • Common coding interview questions
   • Time-complexity optimization challenges
   • Problem-solving scenarios

Technology Stack

• Frontend: Next.js 14, TypeScript, Chakra UI
• AI Integration: Azure OpenAI GPT-4
• Database: Supabase
• Authentication: Clerk
• State Management: Zustand

Innovative Features

Smart Evaluation System

// Two-phase evaluation approach
const evaluationResponse = await makeOpenAIRequest(
  systemPrompt,
  `Please evaluate this ${language} code:\n\n${code}`
);

const statusResponse = await makeOpenAIRequest(
  "You are a code validator. Respond with ONLY 'PASS' or 'FAIL'.",
  `Does this code meet the requirements?...`
);

Scoring System

• Merit-based point allocation
• 10 points per successful challenge
• Persistent progress tracking
• Global leaderboard ranking

Database Schema

CREATE TABLE completed_challenges (
  user_id TEXT,
  challenge_id TEXT,
  completed_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
  PRIMARY KEY (user_id, challenge_id)
);

Educational Impact

Learning Objectives

1. Improve coding skills through practice
2. Receive AI-guided feedback
3. Learn from mistakes without solution revelation
4. Track progress via points system

User Experience

• Clean, intuitive interface
• Mobile-responsive design
• Real-time evaluation feedback
• Progress visualization

Technical Architecture

Evaluation Flow

1. Code submission
2. Duplicate submission check
3. AI evaluation process
4. Result processing and feedback
5. Points award and leaderboard update

Security Features

• Rate limiting implementation
• Authentication integration
• Input sanitization
• Secure API handling

Challenge Creation Guidelines

Best Practices

1. Clear Descriptions

   • Specify exact requirements
   • Include input/output examples
   • Mention performance expectations

2. Progressive Difficulty

   • Start with basic concepts
   • Gradually increase complexity
   • Include edge cases

3. Educational Value

   • Focus on common patterns
   • Emphasize best practices
   • Include real-world scenarios

Template Structure

{
  id: "unique-identifier",
  title: "Descriptive Title",
  description: "Clear, detailed requirements",
  defaultLanguage: "preferred-language",
  defaultCode: "// Example solution or template",
  initialCode: "// Starting point for users"
}

Future Enhancements

• Challenge category system
• Difficulty rating system
• Custom evaluation criteria
• Challenge creation UI
• Challenge sharing platform

Conclusion

The AI Code Evaluation Game represents a novel approach to coding education, combining artificial intelligence with gamification to create an engaging learning experience. Its extensible challenge system makes it adaptable for various training scenarios, from corporate training to academic education.

---

This project was developed for the Ready Tensor AI Competition, showcasing the potential of AI in educational technology.