למשרה היברידית
תיאור המשרה:
Investigate agentic AI architectures: study how agents orchestrate reasoning, memory,
tools, and APIs, identifying weak points in design and control flows.
Threat research and testing: study prompt injection, jailbreaks, hallucination leverage,
retrieval attacks, and toolchain exploits to expose risks in agentic systems.
Prototype detections and mitigations: create PoCs, design runtime policies, monitoring
frameworks, and isolation strategies to prevent agent or tool compromise.
Translate research to product: partner with engineering and product to ship hardened
defaults, telemetry, and protections.
Publish and represent: share insights through papers, talks, and collaborations, shaping the global conversation on agentic AI security.
Shape the company
5-10+ years in cybersecurity R&D, threat research, or AI/ML security.
Experience or strong curiosity in analyzing AI and agentic systems for vulnerabilities such
as prompt injection, memory abuse, jailbreaks, or privilege escalation.
Solid grounding in application and identity security, and runtime protection.
Familiar with AI and agent frameworks, tool use, prompts, and APIs.
Ability to design rigorous experiments, measure impact, and produce actionable insights.
Clear communicator who connects deep research with product and strategy.
Comfortable with ambiguity, short iteration, and direct collaboration with design partners.
Mission driven, curious, and passionate about making AI safer and more trustworthy.
