8EH Radio ITB

Live site

8EH Radio ITB is a deployed web platform for a campus radio organization. It combines live audio, podcasts, articles, programs, announcer profiles, event forms, media-partner services, and an authenticated editorial dashboard.

The project also includes AI-assisted workflows for public questions and newsroom content production.

What this project demonstrates

• Full-stack product development with the Next.js App Router.
• Streaming AI responses for a public radio chatbot.
• Dynamic context assembly from recent articles and podcasts.
• Role-protected content tools for drafting, editing, translation, summaries, and tags.
• Prisma data modeling with MongoDB.
• Authentication, media workflows, event forms, exports, short links, and analytics.

AI and content pipeline

The platform includes an AI chatbot for public questions and role-protected editorial tools for newsroom content production.

• Vercel AI SDK integration with Groq and Google models;
• streaming public Q&A with radio-specific knowledge and recent database content;
• inference endpoints for article drafts, outlines, title ideas, editing, translation, summarization, and tags; and
• rate limiting for public AI endpoints and role checks for editorial tools.

AI pipeline

Visitor question
       |
       v
Next.js API route
       |
       +-- rate limit by client IP
       +-- load recent articles and podcasts
       +-- add 8EH Radio knowledge context
       v
Groq streaming response
       |
       v
Chat widget in the public website

Editorial AI tools use authenticated role checks and Google Gemini for actions such as:

• article title generation;
• article outlines and first drafts;
• editing and translation;
• SEO-style summaries; and
• tag suggestions.

Main routes

Method	Route	Purpose
POST	/api/ai/chat	Public streaming Q&A with dynamic content context
POST	/api/ai/tldr	Public article TL;DR generation
POST	/api/ai/blog	Authenticated title, outline, draft, edit, and translation tools
POST	/api/ai/summarize	Authenticated article description generation
POST	/api/ai/tags	Authenticated article tag suggestions
GET, POST	/api/blog	Read and create blog content with role checks
GET, POST	/api/events	Manage events and form workflows

Architecture

Next.js 15 and React 19
  public radio site, dashboard, chat widget, forms, media pages
          |
          +-- Vercel AI SDK
          |     +-- Groq
          |     +-- Google Gemini
          |
          +-- NextAuth Google OAuth
          |
          +-- Prisma ORM
                |
                +-- MongoDB

Technology

• Frontend: Next.js 15, React 19, JavaScript, TypeScript, Tailwind CSS, SWR
• AI: Vercel AI SDK, Groq, Google Gemini, streamed responses
• Backend: Next.js route handlers, NextAuth, Prisma
• Data and storage: MongoDB, Cloudflare R2 support, audio and image assets
• Deployment: Vercel-compatible build with bun run build

Run locally

Requirements:

• Bun 1.3 or newer
• MongoDB
• Google OAuth credentials for authenticated dashboard features

Install dependencies and start the development server:

bun install
bun dev

Open http://localhost:3000.

Create local environment variables for the features you use:

MONGODB_URL=mongodb://localhost:27017/8eh-radio
NEXTAUTH_SECRET=replace-with-a-local-secret
GOOGLE_CLIENT_ID=replace-with-google-client-id
GOOGLE_CLIENT_SECRET=replace-with-google-client-secret
GROQ_API_KEY=replace-with-groq-key
GOOGLE_GENERATIVE_AI_API_KEY=replace-with-google-ai-key

Media uploads, radio streaming, and production deployment require additional storage and stream configuration. Never commit secrets to the repository.

Useful commands

bun dev
bun run build
bun run lint
bun run format

Current limitations

• Public AI rate limits are stored in memory, so production deployments should use a shared limiter such as Redis.
• AI-generated content should be reviewed by an editor before publication.
• Media uploads and stream configuration require deployment-specific credentials.
• Authentication, storage, and content data should be configured with production privacy and retention policies.

License

This project is released under the MIT License.