Project Overview

Jiko is an AI-powered assistant that monitors your activity and messages you when you start doomscrolling. It uses local device analytics and behavior modeling to nudge you back toward focus, like getting a text from a friend asking if you've studied for your midterm yet. (Work in progress!)

Features:

• Detects when social media or entertainment apps are opened for extended periods
• Considers upcoming homework, exams, papers due on Canvas
• Sends personalized reminders via SMS using Twilio integration
• Analyzes screen time and app usage trends to generate weekly reports
• Includes an adjustable sensitivity system that learns when to intervene without getting blocked

Technologies Used:

• Swift + SwiftUI for macOS and iOS agents that track app usage
• Used Canvas LMS API to track upcoming homework, papers, exams, and other scheduled events.
• Twilio API for real-time SMS notifications

Achievements:

• Built a functional macOS menu bar tracker and iOS app that detect active apps in real time
• Implemented automatic message dispatch through Twilio when distraction thresholds are crossed
• Integrated a personalized chatbot trained on my own text history
• Created early UI mockups (on a piece of paper) for iOS companion dashboard showing focus metrics and screen time breakdown

Project Overview

Mirage is an AI assistant application featuring a MacOS UI overlay that lets users prompt any LLM available through the OpenRouter API.

Features:

• MacOS UI overlay for a native desktop experience
• Supports any LLM available via OpenRouter API
• Node.js backend with RAG (Retrieval-Augmented Generation) and vector search
• Fast, context-aware AI responses for complex queries
• Multi-sectioned architecture for modularity and scalability
• Cloud storage enables cross-platform usage

Technologies Used:

• SwiftUI for MacOS UI overlay
• React with TypeScript for the static landing page
• Hosted Dockerized AWS EC2 for backend and vector embedding
• Node.js backend implementing RAG and vector search
• OpenRouter API for LLM access and AI responses
• AWS EBS for vector data and AWS RDS for chat logs

Achievements:

• Enabled flexible AI assistant experience supporting multiple LLMs
• Created a lightweight MacOS overlay, aimed to increase usability and productivity.

Project Overview

A cross-platform workout tracker and planner designed to help users log and track long-term strength training progress efficiently.

Features:

• Log workouts and track strength progression over time
• Scraped exercise and muscle group data using Playwright
• Cross-platform support for mobile devices via React Native
• User research-informed UI/UX for intuitive day-to-day use

Technologies Used:

• React Native for cross-platform mobile development
• TypeScript for type safety and maintainable code
• Playwright for web scraping and data collection
• Local storage / SQLite for offline use

Achievements:

• Created a dataset of exercises and targeted muscle groups for recommendation engine
• Validated app features through informal student user research

Project Overview

PlanUMN is a comprehensive graduation planning tool designed to simplify course scheduling and degree tracking for University of Minnesota students.

Features:

• Drag-and-drop interactive schedule builder
• Autocomplete course search with real-time suggestions
• Import transcripts to fill classes from previous semesters
• Cloud storage for plans, enabling plan sharing

Technologies Used:

• React with TypeScript for frontend development
• Next.js (serverless) for backend API and SSR/SSG
• Utilized Supabase for relational data storage and user authentication.
• Ensured plans are kept secure with RLS, user authentication, and server-side permission checks.
• RESTful APIs for fuzzy search, quick course info, and manipulating plans.

Achievements:

• Used by over 80 students pre-release

Project Overview

This website serves as my personal portfolio, designed to showcase my projects, experience, and interests in a clean and interactive way.

Features:

• Fully responsive design for desktop and mobile
• Interactive project section with smooth animations
• Custom cursor-reactive gradient effects
• Minimalist design centered on readability and accessibility

Technologies Used:

• Next.js with React and TypeScript
• Tailwind CSS for styling
• Deployed on Vercel for fast global hosting