GitHub Repository: https://github.com/MayankParashar28/Lumina
Live Demo: https://lluminaa.vercel.app
Developer: Mahir (Mayank Parashar)
Project Overview
Lumina is a full-stack intelligent blogging platform that integrates Artificial Intelligence and semantic search to enhance how users create, discover, and interact with content. The platform transforms traditional blogging into a context-aware knowledge ecosystem using vector embeddings and AI-powered analysis.
This project demonstrates strong capabilities in modern web development, backend architecture, API integrations, and scalable system design. Lumina was built with production readiness, performance optimization, and maintainability in mind.
Key Features
AI-Powered Semantic Search
- Context-aware search powered by Google Gemini embeddings
- Cosine similarity matching for accurate content retrieval
- Smart recommendations based on semantic relevance
Intelligent Content Processing
- Automated content summarization
- Related article discovery using embeddings
- Improved search ranking and relevance
Real-Time Experience
- Live notifications using Socket.IO
- Infinite scrolling feed for optimized UX
- Responsive layout across devices
Security and Access Control
- Secure authentication using OAuth and password hashing
- Role-based access control
- Admin moderation dashboard
Technology Stack
| Layer | Technologies |
|---|---|
| Frontend | EJS Templates, HTML, CSS |
| Backend | Node.js, Express.js |
| Database | MongoDB Atlas |
| AI Services | Google Gemini API |
| Real-Time | Socket.IO |
| Deployment | Vercel, Docker |
| Version Control | GitHub |
Architecture Overview
Lumina follows a modular and scalable architecture:
- Client Layer: User interface rendering and user interaction
- Application Layer: Routing, middleware, authentication
- Business Logic Layer: Blog services, AI processing, search engine
- External Services: Gemini API, MongoDB Atlas, CDN
This separation of concerns improves maintainability, scalability, and development velocity.
Engineering Highlights
- Implemented semantic search pipelines using vector embeddings
- Designed secure authentication and authorization workflows
- Built real-time communication channels using WebSockets
- Optimized backend performance with modular service architecture
- Containerized the application for scalable deployments
Learning Outcomes
- Practical experience integrating AI APIs into production systems
- Full-stack development with real-world scalability considerations
- Secure system design and data handling
- Performance optimization and system modularization
- Deployment automation and cloud hosting workflows
Repository and Demo
- Source Code: https://github.com/MayankParashar28/Lumina
- Live Application: https://lluminaa.vercel.app
Career Objective
I am actively seeking internship opportunities in Software Engineering, Full Stack Development, and AI-driven applications where I can contribute to impactful products while continuing to strengthen my engineering skills.
Top comments (5)
Cool project overall. The AI-powered search and recommendation system is definitely the strongest part, and the way you’ve used embeddings for semantic discovery is solid. The backend structure and real-time features look well thought out too. I think the UI could use a bit more polish, and adding a simple architecture diagram or performance insights would make the project even clearer. Still, this is a strong build with real-world potential. Nice work.
THANK YOU FOR YOUR FEEDBACK
Cool. But the cursor on the website is slow (I'm talking about the too low interpolation coefficient).
I'LL TRY TO IMPROVE IT,SIR
Good luck!