Security Researcher — AI & LLM Red Teaming (SouqSecure2024)
Designed and executed adversarial evaluations of LLMs, agents, and RAG pipelines using crafted attack prompts and automated jailbreak/probing campaigns. Built systematic test suites with garak and PyRIT and scored model behaviors using OWASP LLM Top 10 rubrics. Documented reproducible findings with risk ratings and proposed mitigations. • Automated adversarial prompt and jailbreak testing • Evaluated function-calling/tool-use and RAG pipeline vulnerabilities • Scored behaviors against OWASP LLM Top 10 with rubrics • Authored reproducible security reports with mitigations