AIJudge

Problem Statement

AI Judge is an automated hackathon judging platform that uses AI to evaluate projects with consistency and fairness. It streamlines the entire hackathon judging workflow—from project submission to reward distribution—eliminating human bias and reducing the workload for organizers.No more waiting days for results or dealing with inconsistent judging criteria. AI Judge analyzes project descriptions, code repositories, and demo videos to provide comprehensive evaluations based on customizable rubrics, then distributes rewards directly on-chain.🔧How It WorksProject Collection- Add projects via direct URLs or by scanning showcase pagesAI Analysis- Our system extracts and analyzes:Project descriptions and documentationGitHub repositories and code qualityDemo videos (with automatic transcription)Multi-Model Evaluation- Projects are judged by multiple AI models using customizable rubricsTransparent Scoring- Detailed breakdowns show exactly how scores were calculatedOn-Chain Rewards- Distribute POLYGON rewards to winners with a few clicks🔗Key FeaturesAutomated Scraping: Extract project details from showcase pagesVideo Analysis: Transcribe and analyze demo videosMulti-AI Judging: Use multiple models for balanced evaluationCustomizable Rubrics: Adjust criteria weights to match hackathon goalsOn-Chain Rewards: Distribute prizes directly through Polygon💡Why It MattersHackathon judging is often subjective, time-consuming, and prone to bias. AI Judge brings transparency and efficiency to the process, allowing organizers to focus on community building while ensuring fair evaluations. It's like having a team of expert judges working 24/7 with perfect consistency.

Solution

AI Judge combines AI evaluation capabilities with blockchain technology to create an end-to-end hackathon judging and reward distribution system:🧠1. Data Collection & ProcessingWeb Scraping Engine: Built with BeautifulSoup and requests to extract project details from showcase pagesVideo Processing: Uses yt-dlp and moviepy to download and process demo videosTranscription Pipeline: Converts video content to text for AI analysisGitHub Integration: Extracts and analyzes README files and repository information🤖2. AI Evaluation SystemMulti-Model Architecture: Leverages both OpenAI and Anthropic models for balanced judgingCustomizable Rubrics: Flexible scoring system with adjustable weights for different criteriaPrompt Engineering: Carefully crafted prompts ensure consistent and fair evaluationsResult Aggregation: Combines scores from multiple models for more reliable results📊3. User InterfaceStreamlit Frontend: Clean, responsive interface built with StreamlitInteractive Dashboard: Real-time updates and detailed result breakdownsCustomization Controls: Adjust rubric weights and evaluation parametersExport Capabilities: Download results in various formats💰4. Blockchain IntegrationWeb3 Integration: Useswto interact with Polygon networkSmart Transaction Handling: Manages gas estimation and transaction signingBatch Processing: Efficiently distributes rewards to multiple winnersSecurity First: Private keys stored securely in environment variables🔄5. End-to-End FlowOrganizer adds projects via URLs or showcase page scanningSystem extracts all relevant project dataAI models evaluate projects based on customized rubricResults are displayed with detailed breakdownsRewards are distributed on-chain to winning projectsThe system is designed to be modular and extensible, allowing for easy addition of new AI models, evaluation criteria, or blockchain networks in the future.