Senior Staff Security Engineer, AI Safeguards - Moveworks

ServiceNow•Mountain View, CA

15h•$201,300 - $352,300

About The Position

Moveworks, part of ServiceNow, is seeking a Senior Staff Security Engineer focused on AI safeguards. The role maintains a current map of machine-learning usage and safeguards across products and leads security design for new ML features. It tests safeguards against jailbreaks, prompt injection, prompt leaks, and toxicity, then converts findings into evaluation cases that run whenever models, prompts, or safeguards change. The engineer works with ML engineers to design and build mitigations that remain effective even when detection fails, defines and reports security metrics, reviews changes that could weaken safeguards, and tracks AI attack research to develop new tests and mitigations. The position calls for extensive security engineering experience, hands-on adversarial testing of LLM systems, strong knowledge of LLM attacks and defenses, and the ability to implement production fixes. It also requires coding proficiency in Python or Go and enough ML knowledge to build evaluations and assess classifier effectiveness. Compensation includes a base salary of $201,300–$352,300 for positions in the specified location, plus equity when applicable, variable or incentive compensation, and benefits. Actual compensation may vary based on qualifications, skills, competencies, and work location.

Requirements

Typically 8 or more years of experience in security engineering, including 2 or more years focused on AI or LLM security.
Hands-on adversarial experience, such as red teaming, penetration testing, or vulnerability research, applied to LLM systems.
Deep knowledge of LLM attacks and defenses, such as the OWASP Top 10 for LLM Applications and MITRE ATLAS.
Ability to identify risks, develop mitigation plans, and implement fixes personally.
Strong coding skills in Python or Go, with a record of shipping production code, not only findings.
Sufficient ML fundamentals to build evaluations and judge classifier effectiveness.

Responsibilities

Maintain a current map of ML usage and safeguards across products.
Lead the security design of new ML features.
Test safeguards against jailbreaks, prompt injection, prompt leaks, and toxicity.
Turn test findings into evaluation cases that run with every model, prompt, and safeguard change.
Design and build mitigations with ML engineers that remain effective when detection misses.
Define security metrics for safeguard effectiveness and report on them.
Review changes that could weaken a safeguard.
Track AI attack research and convert new techniques into tests and mitigations.