Mohammad AminRoshani
Graduate Research Assistant in the Trustworthy AI Lab, Wayne State University
Generative AI, agentic workflows, and LLM safety in clinical settings.
About
Generative and agentic AI for clinical and biomedical work.
I am a Computer Science PhD student at Wayne State University in the Trustworthy AI Lab with Dr. Dongxiao Zhu. I build language models and agentic systems for clinical and biomedical use. Recent projects include conversational COVID-19 risk assessment, membership inference and LLM privacy, SDOH extraction from MIMIC, and a School of Nursing collaboration on a health-tracking app for older adults.
Education PhD and MS in Computer Science at Wayne State University; BS in Computer Science from the University of Tehran.
Teaching Graduate teaching and instructor roles across programming, web technology, senior capstone, and trustworthy AI.
News
Recent news
Research
Generative AI, LLM safety, and health informatics.
Selected Work
Selected projects
Automatic Calibration for Membership Inference Attacks
Calibration methods for membership inference attacks on large language models.
COVID-19 conversational risk assessment
Generative LLM app for patient and clinician COVID-19 severity risk assessment.
Elderly health tracking app
Health app for older adults with mobility tracking and clinician messaging.
Publications
Recent papers
Automatic Calibration for Membership Inference Attack on Large Language Models
Generative Large Language Model-Powered Conversational AI App for Personalized Risk Assessment: Case Study in COVID-19
Learning to Poison Large Language Models for Downstream Manipulation
Activities
Talks, awards, and teaching
Industry Day on AI
Presenter
Outstanding Graduate Teaching Assistant Award
Award recipient
Senior Capstone Project Presentations
Graduate Teaching Assistant