A personal productivity dashboard on Next.js (App Router) and React: tasks, natural-language scheduling checked against Google Calendar, and lightweight home widgets (clock, greeting, weather, quotes). A LangGraph agent parses intent, reads your calendar, and proposes free slots for overlaps, with human-in-the-loop approval before OAuth writes. IndexedDB captures tasks locally while PostgreSQL and Prisma keep cloud state in sync for the agent.

A modern full-stack e-commerce application with cart and checkout (Stripe), Google OAuth login, and lightweight AI-powered product suggestions using the OpenAI API. Deployed on Vercel, Render, and MongoDB Atlas.

A full-stack vocabulary learning app with flashcards, search, pagination, and animated transitions. Built with MERN stack.

A cloud-hosted bookstore app with React + TypeScript frontend, Java backend on Tomcat, and AWS deployment. Focused on modern UI and accessibility.

Personalized slam book app with user authentication, custom questions, and themed backgrounds. Built using Django and PostgreSQL.

Event management platform with centralized RSVP tracking, Spring Boot backend, and AWS deployment.

Social media clone featuring authentication, posts, comments, image uploads. Built with React and Firebase suite.

Visualization dashboard for hotel booking trends using Dash framework and Google Cloud deployment.

Advanced full-stack app deployment on AWS using custom VPC setup, secure architecture, and Tomcat-MySQL backend.

Scalable big data processing using PySpark on Amazon EMR to analyze large tweet datasets.

Serverless REST API design using AWS Lambda, API Gateway, and RDS MySQL database.

Migration strategy for a clients' web apps from EC2 to a serverless AWS architecture achieving cost optimization.

End-to-end machine learning pipeline deployment for wine quality prediction using AWS SageMaker.

Advanced threat detection using machine learning for URL safety: a multi-class model distinguishes benign, phishing, malware, and defacement links on a large Kaggle malicious-URL dataset (650k+ samples). The pipeline extracts rich structural and statistical features from each URL and evaluates classifiers to support proactive, learning-based detection beyond static blocklists. Implemented in a Jupyter workflow with accompanying project report.