Projects
Autonomous Driving w/ YOLO Algorithm (DEPRECIATED)
With Tensorflow and the YOLO (You Only Look Once) algorithm, I implemented object detection and Non-Max suppression over the COCO dataset.
Project Link: Google Colab Notebook
Art Generation w/ Neural Style Transfer
With Tensorflow, I developed a project that implemented a neural style transfer algorithm that generated novel artistic images. Using a style cost function and a content cost function, the model started off with a random image matrix and iteratively optimized it. This project is based on the VGG network from the original NST paper.
 |  |  |
| Input image 1 | Input image 2 | Output image |
Project Link: Google Colab Notebook
Jazz Music Generation Project w/ LSTM
With TensorFlow, I developed a project that implemented an LSTM sequential model to generate a basic novel Jazz musical clip from training data. Using the gated architecture of Long Short-Term Memory networks, the model captured long-range dependencies in musical sequences, enabling note transitions and rhythmic consistency across the output.
Generated Music Clip
Project Link: Google Colab Notebook
Published Research
NeuroLimbAI: Enhancing Sensory Feedback in an Origami Inspired Prosthetic Arm with Electroencephalogram-Controlled Noninvasive Vibrotactile Haptic Feedback
NeuroLimbAI is a trained Transformer-based neural network that translates non-invasive EEG brain signals into real-time, multi-degree-of-freedom prosthetic arm movements with integrated haptic feedback. I designed an end-to-end pipeline that processed raw EEG data through signal preprocessing and feature extraction stages before mapping neural activity to continuous prosthetic control commands, enabling closed-loop inference. Invited & presented research at Østfold University College in Halden, Norway.
Published in IEEE Xplore, ISBN: 979-8-3503-4977-1
Youtube Link: Video
News Article: Link
VertiPaw: Development of a Vertical Climbing Robot with Adaptive Suction and Suspension Foot Design for Infrastructure Monitoring
VertiPaw is a quadrupedal robotic system developed for infrastructure inspection in hazardous and hard-to-reach environments, capable of vertical climbing and load-bearing using a vacuum-based adhesion mechanism. Experimental results demonstrate stable horizontal and vertical locomotion exceeding target velocities, effective load carrying up to motor torque limits, and reliable surface adhesion across multiple materials. Invited & presented research at Hitachi’s Research Lab in Tokyo, Japan.
Published in IEEE Xplore, ISBN: 979-8-3315-2156-1