DevScan: AI-Powered Hackathon Project Analysis

Inspiration

During hackathon judging, evaluators face a challenge: they need to quickly understand and assess dozens of complex projects within a limited timeframe. We wanted to create a tool that would help judges, mentors, and other hackathon participants efficiently analyze projects by extracting key technical insights and presenting them in an accessible format.

What It Does

DevScan is a web application that uses AI to analyze hackathon projects from DevPost and their associated GitHub repositories. It provides:

Automated Technical Analysis: Extracts key features, implementation details, and complexity assessments
Project Scoring: Evaluates projects based on technical merit, complexity, and feature completeness
Advanced Search: Find projects by keywords, technologies, or specific criteria
Visual Insights: Clean, intuitive UI that presents technical information in digestible formats

To start just replace the "https://devpost.com" => "https://devscan.club" and view the amazing analysis

How We Built It

DevScan is built with a modern tech stack focused on performance and scalability:

Next.js 14: For server-side rendering and optimal performance
React 19: For building the responsive UI components
MongoDB: For storing project data and analysis results
Google Gemini AI: For natural language processing and technical analysis
GitHub API: For accessing repository content and code analysis
TailwindCSS: For beautiful, responsive styling

The architecture follows a modular approach with clear separation of concerns:

Server components for data fetching and initial rendering
Client components for interactivity
API endpoints for external integrations
Dedicated services for analysis operations

Challenges We Ran Into

Rate Limiting: Managing API rate limits across GitHub and Gemini API
Data Extraction: Reliably extracting structured data from unstructured project descriptions
*Performance Optimization *: Ensuring fast search and analysis operations even with complex data
Search Relevance: Creating a search algorithm that returns truly relevant results
Hydration Errors: Resolving React hydration mismatches between server and client rendering

Accomplishments That We're Proud Of

Built a comprehensive project search system with filtering and pagination
Created a sophisticated AI analysis pipeline that extracts meaningful technical insights
Implemented a responsive, accessible UI that works across device sizes
Achieved high performance with server-side rendering and optimized database queries
Developed comprehensive documentation for future contributors

What We Learned

Advanced Next.js patterns for mixing server and client components
Techniques for efficient MongoDB text searching and indexing
Strategies for optimizing large language model prompts
Methods for handling asynchronous operations in React applications
Best practices for TypeScript error handling and type safety

What's Next for DevScan

Machine Learning Model: Train a custom model on hackathon project data to improve analysis accuracy
Team Analysis: Add functionality to analyze team composition and contributions
Real-time Collaboration: Enable judges to share notes and evaluations
Integration with MLH: Connect directly with Major League Hacking's platform
Mobile App: Develop a companion mobile application for on-the-go analysis