My Projects

Go Back

Robotics Project

Read more here!

As part of Ohio State's Fundamentals of Engineering Honors curriculum, I worked on a group robotics project coded in C++ to run on the FEH Proteus. We designed a robot to maneuver a diner-themed course, completing tasks like flipping a burger and changing the jukebox song, using a local positioning system, shaft encoders, and multiple servos. Our robot, assembled from factory parts within a $160 budget, placed in the top 16 out of 65 teams, and the project was a collaborative effort with my talented teammates, supported by instructional staff and our dedicated instructor, Paul Clingan.

The SolidWorks 3d rendering of the robot.

SOLIDWORKS rendering of our robot, SCRAP-E in its final form.

The SolidWorks 3d rendering of the robot.

3D Exploded drawing of SCRAP-E

Connect 4

Check it out here!

Wrote a Connect 4 solver using C++ for the backend and vanilla JavaScript for the frontend, compiled to WebAssembly via Emscripten. Initially, I used a 2-D array and minimax algorithm but later optimized it with a bitboard representation and negamax solver, significantly improving performance. The solver can process thousands of moves per millisecond, providing a strong AI opponent, and is available on my GitHub page.

A connect 4 draw

My connect 4 AI playing against me, resulting in a draw-game

Winding Roads

Check it out here!

I created a personal project to explore wave function collapse algorithms, resulting in an nxn grid of tiles that fit together like a puzzle. The algorithm uses 5 images in 12 total orientations to generate the grid. You can see examples from a tileset I designed on my GitHub page.

A mosaic generated by the wave function collapse algorithm

A mosaic generated by the wave function collapse algorithm

Pro-Con Bot

Check it out here!

This AI-Powered decision maker uses ChatGPT to make a list of pros and cons related to any decision you may have. Just type in a description and you will get a list of pros and cons, as well as detailed insights and questions to help guide your decision making process. *Note that this requires a paid OpenAI plan. Each query costs around 6 cents. You will be prompted to enter an OpenAI API key the first time you use the website.

The Pro-Con Bot website

The homepage of Pro-Con-Bot

Neural Network

Read more here!

My first dive into deep learning! I built a feed-forward neural network using only Numpy Arrays. I trained it on the MNIST handwritten digits dataset as well as Google's QuickDraw! dataset. The doodle network has been trained to recognize airplanes, tornados, bananas, anvils, doors, apples, cats, brooms, cacti, couches, and fish. The digit network can recognize the numbers 0-9

The network recognizing an apple

A brief showcase of the neural network and the code behind it