Projects

InternshipCapstoneCySERHomelab

Python Kamiak HPC SLURM FOA Attack M-Attack V-Attack

About

As part of WSU’s CySER research program in Spring 2026, our team studied how images can be subtly altered to fool AI models into giving wrong or harmful responses. My role was to test different attack methods — specifically FOA Attack, M-Attack, and V-Attack — run them on WSU’s supercomputer (Kamiak), and produce results for the rest of the team. FOA was chosen as our primary method and generated the adversarial images the team then analyzed. The rest of the team worked with those images and put together the research poster.

Poster

Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment — WSU CySER Spring 2026

* FOA Attack is an external repository — not my code. I worked with it as part of this research.