Undergraduate Researcher at AUB. I worked on privacy-preserving AI systems for enterprise LLM applications, focusing on protecting sensitive information before prompts reach language models.
I developed DomainShield, a benchmark and evaluation framework for sensitive-entity masking, and explored retrieval-augmented techniques to improve privacy while preserving task performance across domain-specific enterprise prompts.
The project culminated in a research paper on LLM security and privacy-preserving prompt processing, where I served as a co-author.