Built a Python deployment automation tool integrating Vercel CI and IBM Cloud Code Engine, reducing deployment time by 20% for developers by eliminating repetitive ibmcloud CLI configuration.

Shipped a production React application using Shopify SDKs and live product APIs to integrate commerce data into gameplay. Selected from 150+ participants; deployed and maintained on the Shopify Shop app.

Developed a custom UDP file transfer protocol with stop-and-wait ARQ and error simulation; collected and visualized 8+ performance metrics including throughput vs. error rate using data plots and comparative analysis.

Developed and deployed a golf training tool, ForeSights, leveraging YOLOv5 for golf club detection with 98.0% accuracy and an Intel RealSense Depth Camera for 3D tracking; fine-tuned a large language model (LLM) using Cohere to provide real-time golf swing feedback. Led backend deployment using Google Cloud Platform and ngrok to host the web app, incorporating WebSockets for real-time data updates and enabling seamless integration of computer vision, machine learning, and user interface features.

Deployed an ATS resume scanning tool hosted on Vercel and the backend managed in a Docker container on Google Cloud. Utilized NLP techniques by leveraging the SpaCy en_core_web_sm pipeline to break down and tokenize text, extracting and comparing keywords against industry standards.

Developed a real-time detection system for impaired driving using YOLOv8 and live video feed classification. Created React dashboard to stream 6+ camera feeds for law enforcement; recognized by the Mayor of Kingston. Winning submission at QHacks '24.

Designed an AI-powered mobile stylist using LangChain and SQL to generate outfits from a user's closet. Achieved 95% recommendation accuracy; implemented natural language prompts for real-time feedback. Designed Dilliad, a fashion app offering personalized outfit suggestions, leading to a waitlist of 40 users and receiving first place at QTMA x McKinsey Pitch Day 2024.

Led development of a Chrome extension for "Scholar Pups," integrating Retrieval-Augmented Generation (RAG) using Llama 3.1 and LangChain to provide personalized scholarship application support; designed to simulate the guidance typically accessible to more privileged students by analyzing user inputs and matching them with relevant scholarships and AI-generated writing suggestions based on real application criteria.

Developed an Excel program with arithmetic operations, formulas, and dependencies. Formulated data structures and algorithms including graphs, circular queues, and linked lists to optimize Excel formula and dependency processing.

Collected, cleaned, and classified walking-running data to create an AI classifier using a regression model using SciKit-Learn. Designed a graphical user interface using TKinter to allow users to analyze their data.

Created a light switch attachment that used an Arduino MKR to mechanically turn off and on a light switch using Alexa voice commands. Pivoted from a React app to Arduino programming, utilizing API calls within the Arduino IoT Cloud service, and integrated Alexa functionality through cloud integration. Created a contest website allowing users to input contest details, with real-time updates to a connected noSQL database.