AI Powered Hackathon Judge Assistant

Hackathons receive thousands of submissions, making it impossible for judges. We present a Hackathon Judge Assistant, which uses Azure Cognitive Services and Gemini AI to analyze project submissions.

Comment

AI-Powered Hackathon Judge Assistant

Inspiration

Hackathons receive hundreds of submissions, making it difficult for judges to review each project fairly and efficiently. Manual evaluation often leads to:

  • Decision fatigue, reducing accuracy over time.
  • Inconsistent scoring, leading to potential bias.
  • Challenges in processing multimedia submissions, such as images and speech.

Our goal was to streamline and automate the judging process using AI-powered analysis to ensure fairness, accuracy, and efficiency.

What It Does

Our AI-powered assistant helps judges evaluate projects faster and more effectively by:
✅ Extracting text from project documents and images using Azure OCR.
✅ Analyzing images to detect UI/UX elements and architecture diagrams.
✅ Moderating images to ensure content compliance.
✅ Detecting faces to verify team presence.
✅ Providing AI-powered insights based on hackathon criteria.
✅ Allowing judges to chat with an AI assistant (Gemini AI) to ask about project details.
✅ Generating speech output for verbal project evaluations.

How We Built It

We developed a Flask backend integrated with Azure AI services and Gemini AI:

  • Azure Computer Vision for OCR, image analysis, and content moderation.
  • Azure Face API for face detection.
  • Google Gemini AI for context-aware project evaluation.
  • Azure Speech API for speech synthesis of AI-generated insights.
  • Flask & REST APIs for backend integration.

Challenges We Faced

🔹 Handling multiple file types (PDFs, images, text) required integrating multiple Azure services.
🔹 Maintaining speed and efficiency while processing large-scale hackathon submissions.
🔹 Building ML Pipelines – GitHub Copilot was critical in setting up smooth pipelines between different ML models. Since LangChain doesn’t support Google Gemini, we had to find alternative approaches. Copilot helped us navigate this by suggesting custom API calls and model routing.

Accomplishments We’re Proud Of

✅ Successfully integrated multiple AI services to create a multi-functional evaluation assistant.
✅ Automated project analysis for text and images.
✅ Built an interactive AI chatbot for real-time insights on project submissions.
✅ Enabled speech synthesis for accessible judging.

What We Learned

🔹 How to leverage Azure AI for real-world automation.
🔹 The importance of context-aware AI in project evaluations.
🔹 Best practices for integrating multiple AI APIs efficiently.
🔹 How to optimize AI processing for large datasets while maintaining speed.

What's Next?

🔹 Automated project scoring based on AI evaluation.
🔹 Speech-to-text support for voice-based project submissions.
🔹 Deeper AI insights into project innovation and feasibility.
🔹 Deploying the system as a SaaS platform for global hackathons.

🏆 Award Categories

1️⃣ Best Use of GitHub Copilot

We used GitHub Copilot extensively to accelerate development, improve code quality, and enhance problem-solving:

✅ Code Acceleration – Copilot suggested optimized Python-Flask implementations, reducing development time.
✅ Improved AI Model Integration – Copilot assisted in integrating Azure Cognitive Services, helping us call Face API, Speech API, and Computer Vision API.
✅ Enhanced Debugging & Documentation – Copilot’s intelligent suggestions helped identify bugs and refine complex logic.
✅ ML Pipelines & Open-Source Integration – Since LangChain doesn’t support Gemini, Copilot helped us build custom AI pipelines and integrate relevant open-source code from GitHub projects.

2️⃣ Best Azure App Integration

Our project is working with Azure’s cloud services, ensuring scalability, high performance, and AI processing:

🔹 Azure Computer Vision API – Analyzes handwritten notes, screenshots, and project documentation.
🔹 Azure Face API – Detects team presence and prevents unfair submissions.
🔹 Azure Speech API – Converts AI-generated feedback into speech, making the judging process more accessible.
🔹 Cloud-based Storage – JSON-based storage ensures efficient data retrieval and processing.

💡 Azure Portal Copilot Assistance

Setting up Azure services for the first time can be complex, but Azure Portal Copilot streamlined the process by:
✔ Guiding the setup of instances and APIs, ensuring a smooth deployment.
✔ Providing real-time suggestions and troubleshooting, making service integration effortless.
✔ Optimizing configurations for best performance, helping us maximize Azure’s AI capabilities.

By leveraging Azure AI-powered services, our app efficiently processes multi-modal data (text, images, voice), making it a scalable and powerful solution for hackathon evaluations.

3️⃣ Best Use of Azure AI

Our multi-modal AI assistant enhances the judging experience by mixing text, images, speech, and AI-driven decision-making:

🎙 Voice & Speech Processing – Judges receive real-time feedback in multiple languages using Azure’s Text-to-Speech API.
📸 Image & Handwriting Recognition – Our AI extracts insights from screenshots, hacknotes, and submission materials.
💬 AI-Powered Conversations – Judges interact with Google Gemini AI for real-time insights.
⚖ Fair & Unbiased Judging – Our Responsible AI approach removes unconscious bias from evaluations.

🔥 Why Our Project Stands Out

✔ Real-World Impact – Automates a critical problem in hackathons, saving judges hours of manual evaluation.
✔ Inclusive & Accessible – Supports multiple languages, making hackathons more inclusive.
✔ AI-Driven Fairness – Ensures objective and unbiased scoring, improving the credibility of hackathon results.
✔ Seamless Azure Integration – Leverages multiple Azure AI services to provide real-time insights and scalability.
✔ GitHub Copilot-Driven Efficiency – Uses AI-assisted development to write clean, optimized code faster.
✔ Azure Portal Copilot Assistance – Simplified cloud service setup, deployment, and troubleshooting.

Built for the Azure AI Hackathon 2025 🚀

Built With

  • content
  • file
  • flask-cloud-services:-azure-cognitive-services
  • google-gemini-ai-apis-used:-azure-computer-vision-api-(ocr
  • image-analysis
  • languages-&-frameworks:-python
  • local
  • moderation
  • storage
Share this project:

Updates

Submission history