American Express is seeking a Principal Engineer to lead, mature, and scale the secure adoption of Generative AI and Agentic AI across the enterprise, focusing on designing, implementing, and operating high-performant security guardrails.
Requirements
- 10+ years in security engineering, including 2+ years hands-on with LLM/GenAI security and hardening production AI applications.
- Demonstrated ability to design, implement, test, and performance-tune guardrails.
- Proven delivery integrating AI security platforms and guardrails in production.
- Solid foundation across cloud/data/app security principles.
- Proficiency with public cloud, API gateways, IAM, DLP, observability, and policy-as-code.
- Strong experience with model risk management practices.
- Continually test open-source Generative AI and Agentic AI security guardrails, including, but not limited to guardrail libraries, content filters and red-teaming frameworks.
Responsibilities
- Deliver Generative AI and Agentic AI security reference architectures and integration patterns.
- Continually test, evaluate, implement and operationalize Generative AI and Agentic AI security guardrails.
- Replace underperforming guardrails with more robust capabilities while minimizing disruption.
- Conduct performance testing benchmarks to ensure guardrails are high performant.
- Enhance observability for Generative AI and Agentic AI guardrail systems.
- Evaluate, implement and operationalize Generative AI and Agentic AI red team assessment capabilities.
- Establish pragmatic controls that support rapid yet secure enablement of Generative AI and Agentic AI use cases.
Other
- Mentor senior engineers and drive engineering excellence across disciplines.
- Stay ahead of emerging AI threat vectors and lead response strategies to evolving adversarial techniques.
- Partner with enterprise GenAI enablement, Cloud engineering, Legal, Privacy and Data risk pillars to translate policy and regulatory needs into measurable engineering controls.
- Ensure seamless production guardrail efficacy, performance and incident management.
- Author and accelerate quality model risk documents to ensure timely completion of model risk governance.
