Computer Science student at Arizona State University focused on both applied and theoretical machine learning.

I'm especially interested in new ways deep learning and self-supervised learning architectures can be adapted to better fit more specific tasks.

A metric that measures how much of the representation learned by separate linear encoders can be retrained by a shared encoder. Analyzed through theoretical derivation and validated on both synthetic and benchmark datasets.

A deep learning approach for assessing speech accuracy, fluency, prosody, and overall pronunciation quality using a phoneme-audio alignment matrix.

Low latency computer vision model using task specific spatial attention to estimate seven affect signals from live video.

Python | PyTorch | TensorFlow | Transformers | FastAPI | Scikit-learn | Hugging Face | SQL | Pandas | NumPy | LaTex | Linux

Exploring encoder shareability beyond the controlled linear setting.