🚀 HackSussex

A hackathon project to classify hate speech using AI, featuring a dashboard website, backend server, and user authentication.

HackSussex is a hackathon project aimed at detecting and classifying hate speech using AI. The project consists of:

• A TensorFlow-based AI model trained to classify hate speech, served via Hugging Face.
• A dashboard website (built with Vite and React) that shows top 10 trending hashtags on Bluesky filled with misinformation.
• A backend server (using Deno and Oak) for authentication, user management, and AI inference.
• User authentication powered by Clerk.
• API documentation following OpenAPI 3.0 standards.

• 🎯 AI-powered hate speech classification using TensorFlow & Hugging Face
• 🔥 Real-time dashboard to view and filter harmful content
• 🌍 Server backend with Deno & Oak for authentication and AI inference
• 🔒 Secure user authentication with Clerk
• 🌚 Fully documented API with Swagger (OpenAPI 3.0)
• 🐻 Containerized deployment with Podman

• AI Model: TensorFlow, Hugging Face
• Dashboard: Vite, React
• Backend: Deno, Oak
• Authentication: Clerk
• API Documentation: Swagger, OpenAPI 3.0
• Containerization: Podman

git clone https://github.com/ss2d22/hacksussex.git
cd hacksussex

1. Navigate to the bluesky_livedata directory:

cd bludsky_livedata

2. Initialize conda environment:

conda env create -f bluesky.yml

3. Set environment variable:

export GEMINI_API_KEY="<your gemini api key>"

4. Run the server:

python sandbox.py

server tbd

1. Navigate to the backend directory:

cd server

2. Create a .env file with the following variables:

PORT=your_port
MONGODB_URI=your_mongodb_uri
HUGGINGFACE_API_KEY=your_huggingface_key
CLERK_API_KEY=your_clerk_api_key

3. Run the backend server:

deno run --allow-net --allow-read --allow-env index.ts

For convenience, you can run everything using Podman:

podman-compose up --build

The API is fully documented and compliant with OpenAPI 3.0 standards. You can view the documentation at:

View API documentation here

For any queries or suggestions, please open an issue in this repository or reach out to me on Discord.