Senior AI Model Security Researcher
Core
Design and execute white-box attacks against advanced AI models, checkpoints, and post-training systems to investigate adversarial behaviors like jailbreaks, reward hacking, and privacy leakage.
Role type
Senior AI Model Security Researcher (Adversarial ML)
Builds
High-fidelity experiments, automated attack suites, regression evaluations, and adversarial training environments.
Domain
Artificial Intelligence Security / Adversarial Machine Learning
Deliverable
production ML models
Required skills
adversarial machine learning, AI security, model evaluation, privacy attacks, offensive security, experiment design, reproducible technical evidence, programming for research-prototyping, large language model testing, reward modeling, interpretability, capability elicitation, data poisoning, model supply-chain security, cyber exploitation, cloud security, identity security, sandboxing, secure evaluation environments
Preferred skills
experience converting research findings into production mitigations and security controls
Responsibilities
Design and execute white-box attacks against AI models and post-training systems; Investigate adversarial behaviors including jailbreaks, reward hacking, and deceptive behavior; Build automated attack suites that run continuously against new configurations; Develop threat models and assess severity and exploitability; Partner with model scientists and security teams to drive verified remediation; Convert confirmed weaknesses into regression evaluations and safety training data; Communicate complex attack paths and residual risk to leadership and reviewers.
