I joined CTG full-time in May 2025, right before graduating from the University of Delaware. Since then, I've been contributing to a multi-platform project with a React Native mobile app, a React web dashboard, and a Spring Boot backend.

I've worked end-to-end across the stack - designing responsive UIs, building mobile features, and implementing scalable backend APIs. Some highlights include integrating OAuth 2.0 PKCE authentication for the web app, refactoring backend logic to improve API performance, and debugging complex UI and state-management issues in the mobile app.

Working here has sharpened my ability to think holistically, as I've been balancing frontend experience, backend architecture, and deployment workflows, all while learning from a highly collaborative and talented team.

During my internship, I helped build Compass, an internal web app for analyzing government contract bid protests by indexing over 800,000 PDFs.

I developed the Elasticsearch-based search engine, implemented SQL filters for related entities like attorneys and agencies, and helped containerize the backend and some Python scripts with Docker for smoother CI/CD.

It was my first experience designing production-grade systems end-to-end; blending full-stack engineering, search optimization, and backend scalability was an amazing learning experience and it laid the groundwork for everything I do now in my full-time engineering position.

A modern calorie and nutrition tracking app built with React Native. Users can log meals manually or upload food images for AI-powered calorie estimation. I used Keras for an initial binary classification of food vs non-food, converted the trained model to TensorFlow.js and exported the tokenizer for web deployment, then piped positive results to the OpenAI Vision API for a calorie estimation. Data syncs in real-time via Firebase. Demo video above.

Built in 24 hours for the 2023 HenHacks Hackathon, Chirp helps students learn sorting algorithms through interactive visualizations and games. Designed for clarity and accessibility, it earned 2nd place in the Educational category out of 150+ competitors.

Designed and trained CNN and RNN models to classify URLs as phishing or legitimate using a dataset of 800K+ samples for a class project with my partner. Implemented a character-level CNN with Conv1D-MaxPooling-Dense layers, dropout regularization, and binary cross-entropy optimization via Adam, achieving 97% accuracy.

The CNN matched RNN performance while delivering significantly faster inference, proving its efficiency for real-time phishing detection.

Here's my portion of the work (the CNN model) , and here's the full research report .

My first personal project, a classic chess game built in Java with a custom Swing GUI. Implements full chess logic while emphasizing object-oriented design principles. Demo video above.