ReQuest

Cross-platform full stack mobile app for scanning items, finding recycling centers, and earning points for recycling.

• Uses Google Cloud Vision and Places APIs to identify items and locate recycling centers nearby.
• Employs a Llama model to assess recyclability and award points, with Firebase tracking user progress.
• Maintains a crowdsourced database of 47 recycling locations, updated via web integration.

Technologies:

React Native

TypeScript

Expo

Tailwind

Llama Model

Firebase

Google Cloud

Promptly

AI-powered platform for web scraping and automation using natural language instructions, with agentic email delivery, and a real-time playground.

• FastAPI backend with MongoDB for storing tasks, results, and user preferences.
• RESTful endpoints for creating, running, viewing, and updating tasks programmatically.
• Dockerized deployment on Render for scalability and portability.

Technologies:

React

TypeScript

Next.js

Python

FastAPI

Docker

MongoDB

Gemma Model

Arcade

Render

PanicPrep AI

AI-powered flashcard generator with account management, real-time storage, and payment integration.

• Firebase for online storage and Clerk for secure user authentication and flashcard management.
• Stripe API for payments and subscriptions to premium features.
• Mixtral model via Groq API generates study content tailored to user preferences and progress.

Technologies:

Firebase

Stripe API

Mixtral Model

Mui

React

Node.js

Next.js

Clerk

SnipDeck

Chrome extension for managing text snippets and screenshots, featuring speech synthesis and local storage.

• Integrates with Chrome Extensions API for seamless browser functionality.
• Uses Speech Synthesis API for reading snippets aloud, enhancing accessibility.
• JavaScript manages context menus, screenshot capture, and WebP image compression.

Technologies:

Chrome Extensions API

Speech Synthesis API

JavaScript

HTML

CSS

Service Workers

Cyclistic Case Study

Case study analyzing 6.1 million rides of Divvy bikeshare data with R for data manipulation, visualization, and analysis.

• Uses tidyverse for data cleaning and manipulation.
• Calculates geographical distances with geosphere to analyze cycling patterns.
• Processes dates using lubridate to uncover time-based trends.
• Creates interactive maps with leaflet to visualize routes and hotspots.

Technologies:

R

tidyverse

leaflet

Data Analysis

S&P 500 Sector Portfolio Optimization

Optimization algorithm using machine learning, Monte Carlo simulations, and the Sharpe Ratio to predict optimal sector weightings.

• Random forest regression predicts optimal sector weights.
• Monte Carlo simulations optimize risk-adjusted returns.
• Pandas prepares and manipulates historical stock data for analysis.

Technologies:

Python

Scikit-learn

Pandas

Jupyter Notebooks

Machine Learning

MIPS Math Memory Match

Memory match game built in MIPS assembly to explore low-level programming and memory management.

• Game logic, user input, and memory management handled entirely in MIPS assembly.
• Features card shuffling, match checking, and score tracking.

Technologies:

MIPS

Assembly

Memory Management

OrgPortal (Work in Progress)

Full stack web app for UTD's 450+ student organizations, supporting member management, engagement tracking, and org discovery.

• Helps orgs manage members, track engagement, and gain insights; acts as a discovery platform for students.
• Currently refining features and preparing for deployment with improved scalability and data integrity.

Technologies: