Staff ML Engineer, AI Safeguards

ServiceNow • Mountain View, CA

1d • Onsite

About The Position

Moveworks, acquired by ServiceNow in December 2025, builds an agentic AI assistant platform that helps employees interact with business systems using natural language. The Staff ML Engineer, AI Safeguards will develop and operate safeguards for AI agents, helping protect against prompt injection, jailbreaks, toxicity, sensitive topics, and exposure of sensitive data. The role involves designing safeguards for the agent harness, evaluating their effectiveness, collaborating with Security, integrating detection models, and monitoring production behavior. The work requires balancing safeguard quality, latency, and cost, and creating configurable policies that let customers determine how sensitive topics are handled. The position is required in office.

Requirements

• Typically 8 or more years of experience building production software, including 3 or more years of applied machine learning experience training, evaluating, and deploying models in production systems.
• Hands-on experience developing LLM safeguards, such as guardrail models, content classifiers, prompt-injection or jailbreak detection, or safety evaluations.
• Strong model-evaluation fundamentals for agent trajectories and text classification, with attention to training- and evaluation-data quality.
• Working knowledge of how LLM safeguards are bypassed, including indirect prompt injection through tools and documents and multi-turn jailbreaks.
• Excellent Python experience.

Responsibilities

• Design, build, and ship agent-harness safeguards against prompt injection, single- and multi-turn jailbreaks, toxicity, sensitive topics, and sensitive data exposure.
• Build evaluations that measure safeguard efficacy for both false positives and true positives, using customer reports, adversarial testing, and production-traffic datasets.
• Collaborate with the Security team to convert adversarial findings into evaluation cases and fixes, and develop mitigations that remain effective when detection misses.
• Select, integrate, and fine-tune detection models for a stack shared by commercial and government environments.
• Balance quality, latency, and cost for safeguards running in the request path.
• Build configurable safeguard policies that allow customers to choose how sensitive topics are handled.
• Monitor safeguard behavior in production and fix regressions.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed