Projects

Pupil Detection Neural Network: CyVision Internship

• Developed a UNET and Mask R-CNN convolutional network to detect pupils in a set of infrared images to track the eye's movement.

• Used augmentation to get over the limited training data and custom loss function to prevent disconnected and non-elliptical regions

• The resulting model excelled with an accuracy of over 99 percent with a dice score of 98.3 percent.

AP Tutor Chatbot

• Fine-tuned AI model knowledgeable about past AP exam questions along with their answers

• Utilizes embedding models to determine the AP and the unit being asked 

• From the embeddings, it pulls up the relevant textbooks and past exams to give high-quality answers

Grow MVP

• Created an MVP for the company I co-founded called Grow, helping authorizes creating prescribed burn plans

• Used matplotlib to mark an area of the map, select the parameters for the tarsecs, and fill out the pdf with AI assistance

Chess Game

• Used PyGame to visualize a chess game with an AI opponent and two-player game mode

• Used min-max algorithm and recursive backtracking to create an AI opponent that can play up to upper-intermediate level