work

At the core, I’m someone who enjoys building and designing - whether that be developing a web tool or creating a usable and thoughtful app design. I try to ensure the technology I build is sustainable, scalable, and made with humans in mind. Here’s where you can take a look at my skills applied for real world impact.

SPRING 2024 V1 INCUBATOR

Claris

A full-stack integrated web application created to help U-M labs keep their materials centralized, data trustworthy, and research moving forward.

Features four features in one: a DOI lookup with attached PDF editor, DropBox and file integration, LLM (AI) assistance trained on lab-specific data, and collaboration capabilities
Key parts of this project included consumer research with more than 7 labs here at U-M, splitting data into vectorized data bases, integrating numerous existing softwares and adapting them to web-based technologies, and maintaining a brand identity as a pre-seed startup.
This project was featured at V1 Demo Day, Michigan’s showcase for student-built projects and ventures.

TypeScript JavaScript Node.js React Supabase Pinecone

FALL 2024 MHACKS 2024

Reguard

Food, drug, and item recalls increased nearly 20% from 2020 to 2023. We saw a need for an easily accessible health & safety app in our community and decided to use technology and software to help the everyday consumer.

ReGuard is a mobile application created in Swift that utilizes Apple Developer APIs (VisionKit & MapKit) to help consumers stay up-to-date on food and drug recalls, addressing a need for public safety and health technology
It integrates OCR, barcode scanning, and map notifier using synced FDA recall data to alert users. Maps and geolocation are used to alert users based on local stores.
This was created for MHacks 2024, a 24-hour hackathon at U-M, where I lead development of the map functionality as well as some of the initial SwiftUI framework.

Swift Apple Developer APIs Objective C SwiftUI Python

SUMMER 2024 TALARIA INSTITUTE

LightCNNRad

A quicker, lightweight convolutional neural network (CNN) for radiology image analysis and tumor detection.

LightCNNRad is a machine learning model, namely a convolutional neural network (CNN) aimed to speed up tumor detection in radiology images. accuracy in a 10-epoch test after extensive optimization
It Achieved 86% in a 10-epoch test after extensive training and optimization. I performed vector and tensor math across a robust data set of 5,127 images, and executed the full pipeline of building a CNN model including batch normalization, introducing ReLU, and forward propagation with RMSProp
Credited on the public repository as a contributor and currently published in ResearchGate

Python Pandas TensorFlow Jupyter Notebook