Projects
🚀 A showcase of my work and the technologies I've used to bring ideas to life.
iSense
Decoding vision, delivering insight
May 2025 – June 2025
iSense is an AI-powered image analysis platform that integrates computer vision and natural language processing. Users can upload images and ask questions in natural language to receive intelligent, contextual answers, powered by the LLaMA 4 Scout model on Groq's high-performance infrastructure.
Key Features
- Image understanding using Groq's LLaMA 4 Scout model
- Conversational Q&A about image content
- Real-time image upload and preview with support for multiple formats
- Secure user authentication through Clerk
- Fully responsive and animated UI built with Tailwind CSS
- Planned support for persistent chat history
Technologies Used
- Next.js 15
- TypeScript
- Tailwind CSS
- Groq AI
- LLaMA 4 Scout
- Clerk
- Lucide React
Tazkz
Not the regular task manager
June 2025
Tazkz is a modern task and note-taking application that reimagines productivity by combining tasks and thoughts in a single, unified interface. Built with Next.js 15 and Tailwind CSS, it provides a fast, clean, and intuitive experience to help users plan, manage, and track their day effectively.
Key Features
- Smart task management with parent-child hierarchy, progress tracking, and priority-based sorting
- Contextual note-taking directly within tasks
- Real-time progress visualization using interactive sliders and indicators
- Secure authentication with protected routes powered by Clerk
- Responsive and animated UI built using shadcn/ui
- Robust full-stack setup using Next.js 15 (App Router), Supabase (PostgreSQL), and TypeScript
Technologies Used
- Next.js 15
- TypeScript
- Tailwind CSS
- Supabase
- PostgreSQL
- Clerk
- shadcn/ui
HackSphere 2025 Landing Page
A responsive landing page for a global hackathon event
May 2025
A responsive landing page for a fictional global hackathon event built with React and TypeScript. This project was created as an internship screening task to demonstrate front-end development skills, featuring modern React with TypeScript and functional components.
Key Features
- Modern React with TypeScript and functional components
- Responsive design that works across mobile, tablet, and desktop
- Clean, modular CSS with separate stylesheets for each component
- Header with navigation links to different sections
- Hero section with striking banner, event name, tagline, and registration CTA
- About section with detailed hackathon description
- FAQ section answering common questions
- Contacts section with email and phone contact information
- Footer with copyright information and social media handles
Technologies Used
- React 19
- TypeScript
- Vite
- CSS3
Course Selling App Backend
Role-based API for course publishing, discovery, and purchases
November 2025
Course Selling App Backend is a production-style REST API that powers a course marketplace with separate user and admin authentication domains. Administrators can create and manage courses, while users can discover offerings, purchase courses, and view their enrolled content through secured endpoints.
Key Features
- Separate JWT authentication flows for users and admins
- Secure signup and signin with bcrypt password hashing
- Admin-protected endpoints for course creation and updates
- Public course preview endpoint for browsing catalog
- User-protected purchase workflow for course enrollment
- Purchased courses retrieval for authenticated users
- Modular Express architecture with dedicated routes and middleware
Technologies Used
- Node.js
- Express.js
- MongoDB
- Mongoose
- JSON Web Token (JWT)
- bcrypt
- dotenv
PortfolioIQ
Privacy-first federated intelligence for smarter portfolio decisions
March 2026
PortfolioIQ is a federated learning-based portfolio rebalancing platform where users upload holdings, generate market-aware features, and train a logistic regression model directly in the browser. The system predicts whether to Rebalance or Hold with confidence scoring, while preserving privacy by sending only model weights to the backend for FedAvg aggregation into a continually improving global model.
Key Features
- Privacy-preserving local model training in the browser using TensorFlow.js
- Portfolio upload and parsing support for both CSV and Excel files
- 12-factor feature engineering pipeline combining portfolio and NIFTY 100 market signals
- Rebalance vs Hold prediction with confidence score and condition insights
- Federated averaging (FedAvg) of user model weights to improve a shared global model
- JWT-based authentication with secure user weight history tracking
- Admin workflow to seed and retrain the global model from dataset and user contributions
Technologies Used
- Next.js 16
- React 19
- TypeScript
- Tailwind CSS 4
- TensorFlow.js
- Express 5
- FastAPI
- scikit-learn
- PostgreSQL
- Drizzle ORM
- Turborepo
- Docker
QuickHand
Role-based marketplace for connecting customers with local handymen
Currently Building
QuickHand is a full-stack platform that helps customers post service requests and enables handymen to discover and accept relevant jobs by category. The product includes secure role-based authentication, dedicated customer and handyman flows, and dashboard-level tracking for pending and completed work. It is currently being built as a scalable monorepo with shared backend, auth, and database packages.
Key Features
- Dual-role authentication for customers and handymen
- Role-specific signup and signin journeys
- Job listing with category and pay range metadata
- Category-filtered job discovery for handymen
- Direct hire browsing grouped by handyman specialty
- Job acceptance flow for converting listings to active jobs
- Dashboard insights for completed and pending jobs
Technologies Used
- Next.js
- React
- TypeScript
- Hono
- Bun
- Drizzle ORM
- PostgreSQL
- Tailwind CSS
- Docker
iSense
Decoding vision, delivering insight
May 2025 – June 2025
iSense is an AI-powered image analysis platform that integrates computer vision and natural language processing. Users can upload images and ask questions in natural language to receive intelligent, contextual answers, powered by the LLaMA 4 Scout model on Groq's high-performance infrastructure.
Key Features
- Image understanding using Groq's LLaMA 4 Scout model
- Conversational Q&A about image content
- Real-time image upload and preview with support for multiple formats
- Secure user authentication through Clerk
- Fully responsive and animated UI built with Tailwind CSS
- Planned support for persistent chat history
Technologies Used
- Next.js 15
- TypeScript
- Tailwind CSS
- Groq AI
- LLaMA 4 Scout
- Clerk
- Lucide React
Tazkz
Not the regular task manager
June 2025
Tazkz is a modern task and note-taking application that reimagines productivity by combining tasks and thoughts in a single, unified interface. Built with Next.js 15 and Tailwind CSS, it provides a fast, clean, and intuitive experience to help users plan, manage, and track their day effectively.
Key Features
- Smart task management with parent-child hierarchy, progress tracking, and priority-based sorting
- Contextual note-taking directly within tasks
- Real-time progress visualization using interactive sliders and indicators
- Secure authentication with protected routes powered by Clerk
- Responsive and animated UI built using shadcn/ui
- Robust full-stack setup using Next.js 15 (App Router), Supabase (PostgreSQL), and TypeScript
Technologies Used
- Next.js 15
- TypeScript
- Tailwind CSS
- Supabase
- PostgreSQL
- Clerk
- shadcn/ui
HackSphere 2025 Landing Page
A responsive landing page for a global hackathon event
May 2025
A responsive landing page for a fictional global hackathon event built with React and TypeScript. This project was created as an internship screening task to demonstrate front-end development skills, featuring modern React with TypeScript and functional components.
Key Features
- Modern React with TypeScript and functional components
- Responsive design that works across mobile, tablet, and desktop
- Clean, modular CSS with separate stylesheets for each component
- Header with navigation links to different sections
- Hero section with striking banner, event name, tagline, and registration CTA
- About section with detailed hackathon description
- FAQ section answering common questions
- Contacts section with email and phone contact information
- Footer with copyright information and social media handles
Technologies Used
- React 19
- TypeScript
- Vite
- CSS3
Course Selling App Backend
Role-based API for course publishing, discovery, and purchases
November 2025
Course Selling App Backend is a production-style REST API that powers a course marketplace with separate user and admin authentication domains. Administrators can create and manage courses, while users can discover offerings, purchase courses, and view their enrolled content through secured endpoints.
Key Features
- Separate JWT authentication flows for users and admins
- Secure signup and signin with bcrypt password hashing
- Admin-protected endpoints for course creation and updates
- Public course preview endpoint for browsing catalog
- User-protected purchase workflow for course enrollment
- Purchased courses retrieval for authenticated users
- Modular Express architecture with dedicated routes and middleware
Technologies Used
- Node.js
- Express.js
- MongoDB
- Mongoose
- JSON Web Token (JWT)
- bcrypt
- dotenv
PortfolioIQ
Privacy-first federated intelligence for smarter portfolio decisions
March 2026
PortfolioIQ is a federated learning-based portfolio rebalancing platform where users upload holdings, generate market-aware features, and train a logistic regression model directly in the browser. The system predicts whether to Rebalance or Hold with confidence scoring, while preserving privacy by sending only model weights to the backend for FedAvg aggregation into a continually improving global model.
Key Features
- Privacy-preserving local model training in the browser using TensorFlow.js
- Portfolio upload and parsing support for both CSV and Excel files
- 12-factor feature engineering pipeline combining portfolio and NIFTY 100 market signals
- Rebalance vs Hold prediction with confidence score and condition insights
- Federated averaging (FedAvg) of user model weights to improve a shared global model
- JWT-based authentication with secure user weight history tracking
- Admin workflow to seed and retrain the global model from dataset and user contributions
Technologies Used
- Next.js 16
- React 19
- TypeScript
- Tailwind CSS 4
- TensorFlow.js
- Express 5
- FastAPI
- scikit-learn
- PostgreSQL
- Drizzle ORM
- Turborepo
- Docker
QuickHand
Role-based marketplace for connecting customers with local handymen
Currently Building
QuickHand is a full-stack platform that helps customers post service requests and enables handymen to discover and accept relevant jobs by category. The product includes secure role-based authentication, dedicated customer and handyman flows, and dashboard-level tracking for pending and completed work. It is currently being built as a scalable monorepo with shared backend, auth, and database packages.
Key Features
- Dual-role authentication for customers and handymen
- Role-specific signup and signin journeys
- Job listing with category and pay range metadata
- Category-filtered job discovery for handymen
- Direct hire browsing grouped by handyman specialty
- Job acceptance flow for converting listings to active jobs
- Dashboard insights for completed and pending jobs
Technologies Used
- Next.js
- React
- TypeScript
- Hono
- Bun
- Drizzle ORM
- PostgreSQL
- Tailwind CSS
- Docker