Remote job
Forward Deployed Engineer, AI Trust and Reliability
Job details
About this role
Role overview
This forward-deployed engineering role makes AI agents safe and dependable in production environments for enterprise partners. You will tailor safeguards to partner policies and risk, test deployments against misuse and attacks, and connect partner security teams with engineering so issues are resolved before launch and monitored afterward.
Responsibilities
- Configure controls for what an AI agent may do, when it must refuse, and when it should hand work to a person; account for partner rules, regulations, and supported languages. - Protect sensitive data and consequential actions with measures such as masking, identity checks, and customer consent before booking or payment changes. - Red-team deployments for prompt injection, jailbreaks, data exposure, unauthorized actions, and social engineering; convert findings into fixes and regression tests. - Build automated adversarial and release test suites, including functional, safety, load, and regression checks, and define launch approval and rollback criteria. - Lead partner security questionnaires, architecture and model-risk reviews, penetration-test follow-up, and documentation suitable for audits. - Monitor live deployments for reliability, abuse, and safety failures; coordinate incident response and communicate clearly with partners. - Establish secure partner data access and handling practices, including storage, separation, retention, deletion, and use in model training.
Requirements
- At least five years in security engineering, application security, AI safety, quality engineering, or site reliability, with hands-on production systems experience. - Practical experience red-teaming language-model applications or agents and closing discovered security gaps. - Experience designing and tuning safeguards for an AI product while balancing risk controls with legitimate customer use. - Background representing a technology provider in enterprise security reviews and resolving penetration-test findings. - Ownership of production testing and release quality, including automation, release gates, load testing, and rollback planning. - Strong Python skills for building test harnesses and tooling; ability to explain technical risk to engineers, security teams, and executives. - Experience with travel or financial-services systems and associated fraud, account takeover, or payment risks is sought.
Nice to have
Experience with secure on-premise partner integrations and multilingual testing would support the deployment work described.