Projects

Cornell Autonomous Sailboat Team (CUSail)  

Full Team Lead · Sep 2023 – Present

I lead a 40-member student team developing an autonomous robotic sailboat. My role involves directing software, mechanical, electrical, and business divisions, running design reviews, and coordinating development cycles. I spearheaded the transition to a microservice-based architecture with ROS 2, building scalable APIs and modular robotics software. I also developed a Flutter app (deployed via Firebase) for remote control and a MERN-based telemetry server with REST APIs, routing HTTP/TCP connections using Ngrok.

Repos:
GitHub – sailbot
GitHub – controller app

Decentralized Voting for Prescription Validation

OCaml, Dune, OUnit, Blockchain · Fall 2024

Built a full-stack decentralized validation platform using OCaml and a custom hybrid Proof-of-Work/Proof-of-Authority blockchain. I developed authentication, task assignment logic, and a mining framework, coordinating execution via Dune across 8 modular compilation units. Achieved over 91% code coverage using OUnit and Bisect_ppx. Persistent state was stored in JSON and CSV, simulating real-world prescription consensus scenarios.

Repo:
GitHub – cs3110-final-proj

Crazy Crush – 2D Infinite Runner Game

Unity, C# · Spring 2024

Built in 24 hours during the 2024 DiGA Game Jam, this Valentine’s-themed infinite runner game features a player fleeing their “crazy crush” through an obstacle course. Worked on core gameplay mechanics and sprite-based animation. The game is an ongoing project with future plans for improved scoring and power-ups.

Repo: GitHub – Crazy Crush

TinyTutor – RAG on a Budget

FAISS, BERT, Streamlit, Docker · Spring 2024

TinyTutor is a lightweight full-stack tutor app built for low-compute environments. I engineered a custom FAISS-based vector database integrated with keyword-augmented BERT embeddings to enhance retrieval quality. The app uses Llama for response generation and is containerized in Docker for deployment. It was selected as a finalist at the 2024 Cornell AI Hackathon.

Neural Networks from First Principles

Octave/MATLAB · Spring 2023

Recreated core machine learning models from scratch, including Linear/Logistic Regression, SVMs, and Deep Neural Networks. Implemented custom forward/backward passes, activation functions, and gradient descent with full matrix-vector operations, without using ML libraries—focusing on mathematical understanding and optimization.