AI Security Engineer

📋 Description

• • Perplexity is at the forefront of revolutionizing information discovery and knowledge access through cutting-edge AI. We are seeking a highly skilled, experienced, and hands-on AI Security Engineer to become an integral part of our dedicated security team. In this pivotal role, you will be instrumental in driving the protection of our next-generation AI systems against a constantly evolving landscape of adversarial threats. Your expertise will be crucial in designing, building, and implementing robust security mechanisms that safeguard our self-hosted models, LLM APIs, agents, MCPs (Model Control Planes), and the core AI infrastructure that powers our innovative products.
• • You will be empowered to enable our development teams by providing them with essential tools, clear guidance, and direct technical contributions. This collaborative approach ensures that innovation can flourish while maintaining the highest standards of AI security, making security a strong default rather than an afterthought. Our dynamic and modern tech stack includes Python, NextJS, TypeScript, Docker, AWS, Kubernetes, and PostgreSQL, providing a rich environment for you to apply and expand your skills.
• • A core responsibility will be to define, build, and continuously refine the mechanisms that secure our complex AI systems. This includes proactively addressing adversarial behavior of all kinds, from subtle manipulation to sophisticated attacks targeting our self-hosted models, LLM APIs, agents, MCPs, and other critical components of the AI stack. You will delve deep into technically complex AI systems, meticulously identifying potential weaknesses in their architecture and implementing effective improvements to fortify them.
• • A significant portion of your time, at least 50%, will be dedicated to hands-on remediation efforts. This involves not only fixing identified vulnerabilities but also working closely and collaboratively with peer engineers across different teams to drive the implementation of these remediations, fostering a culture of shared security responsibility.
• • You will be responsible for planning and executing comprehensive threat modeling activities. This includes conducting realistic threat simulations across all our offerings to proactively uncover potential attack vectors and vulnerabilities before they can be exploited. Your insights will be critical in understanding the threat landscape specific to AI systems.
• • Furthermore, you will conduct rigorous cybersecurity evaluations and lead AI security assessments in a cross-functional environment. This requires working effectively with product, engineering, and other stakeholders to ensure a holistic approach to security.
• • You will develop and champion initiatives that significantly enhance our capabilities in evaluating AI systems. This includes improving our prevention strategies, strengthening our detection mechanisms, refining our response protocols, and advancing our threat hunting expertise within the AI domain.
• • A key aspect of this role involves providing expert guidance and education to our developers. You will help them understand potential threats, implement secure coding practices, and adopt security-first mindsets to deter and prevent threats at the source, empowering them to build secure AI applications.
• • This role offers a unique opportunity to shape the future of AI security at a rapidly growing company. You will have the autonomy to innovate, the support to implement your ideas, and the chance to make a tangible impact on the security posture of advanced AI technologies. Your work will directly contribute to Perplexity's mission of providing reliable, secure, and trustworthy AI-powered information.