Arafat Ansari

I'm the founder at KageX, where we work on AI security / offensive security / LLM safety/ Trainings. We have recently shipped Mirage, an open-source autonomous red-team agent for testing AI systems against prompt injection, adversarial testing and policy bypass. With 7+ years of experience specialising in AI security, LLM red teaming, and adversarial testing infrastructure. I design offensive platforms, labs, and large-scale testing environments to discover how AI systems fail in real-world conditions. Our work focuses on making adversarial AI testing reproducible and evidence-based rather than reliant on models grading themselves.