Currently researching how to create contextually coherent and natural looking adversarial patches in images to throw off computer vision models.
Past work includes use of Graph Neural Networks as meta-classifiers to detect trojan DNN models and LLM-guided label-free open-vocabulary object detection.
During my Masters I worked on Using Text to Improve Classification of Man-Made Objects: Masters Thesis.
I’m always looking to collaborate on Adversarial and Trustworthy ML, and LLM/Agent-guided frameworks.
- Languages:
,
,
,
,
- Machine Learning/Data Science:
,
,
,
,
,
,
,
,
,
,
- Web Technologies:
,
,
,
,
,
,
, SCSS,
- Other Tools and Technologies:
,
,
,
.
- Operating Systems:
,



