AI Cyber Adversarial Researcher - entry level

We are seeking an AI Cyber Adversarial Researcher (Junior) to support cutting-edge penetration testing and red teaming of AI/LLM applications. You’ll work alongside senior offensive researchers to prototype and test novel, vendor-agnostic approaches to exploiting and hardening AI systems, guardrails, and agentic platforms—always in a safe, ethical, and controlled environment. This role is ideal for early-career candidates who bring foundational offensive security skills (including malware and shellcode exposure) and want to apply them at the frontier of AI security.

Key Responsibilities

• Conduct AI–cybersecurity research (e.g., GenAI & shellcode workflows, polymorphic malware) to identify risks and defense gaps.
• Develop and test steganographic and cryptographic techniques for prompt security and LLM control.
• Analyze and simulate malware/shellcode behavior; assess AI’s impact on attacker tradecraft.
• Design automated multi-agent attack-chain experiments and red-team simulations.
• Explore NLP-based obfuscation and defensive countermeasures for IoC evasion.
• Identify vulnerabilities, ethics, and mitigations; produce clear technical reports and PoCs.
• Collaborate with Security, ML, and Product teams to turn research into detections and guardrails.
• Stay current with AI security, malware, cryptography, and model safety trends.

Required Qualifications

• B.Sc. in CS, Cybersecurity, EE/CE, or related field (M.Sc. preferred).
• 3+ years in red teaming, malware analysis, or reverse engineering.
• Proficient in Python; familiar with LLM/agentic tooling (ADK, A2A, MCP) and REST APIs.
• Understanding of exploits, C2, OPSEC, and LLM vulnerabilities (prompt injection, jailbreaks).
• Strong research drive, communication, and documentation skills.
• Based in Israel; able to collaborate across time zones.

Preferred Skills

• Red teaming, CTFs, exploit/malware R&D experience.
• Knowledge of cryptography, steganography, and detection engineering (YARA/Sigma).
• Familiarity with NLP/LLM stacks, vector databases, and containerized ML infra.
• Certifications: OSCP, OSCE, or CRTO.

Ethics & Safety

All research is conducted in controlled environments with strict legal, safety, and policy oversight. Offensive techniques (including malware and shellcode) are used exclusively for defensive research, testing, and improving safeguards.

If you are a forward‑thinking junior researcher in Israel eager to build at the intersection of offensive cyber and AI security, we encourage you to apply.