Red Team Program Manager

📋 Description

• • Gray Swan AI is at the forefront of protecting organizations from the rapidly evolving landscape of AI security threats. We specialize in developing cutting-edge, real-time threat detection systems, automated validation tools, and adaptive defense mechanisms specifically designed for AI labs and enterprises. As a rapidly growing, well-funded company with a lean team of 25, we are seeking a dynamic and experienced Red Team Program Manager to join our mission.
• • In this pivotal role, you will be instrumental in orchestrating the design, launch, and ongoing success of our AI Red Team program. This includes coordinating the development of our Arena competitions, managing private red-teaming engagements with leading AI labs and businesses, driving the recruitment of top-tier red team talent, and contributing to the continuous improvement of our Arena platform. You will work at the intersection of AI safety, security, and product development, collaborating with some of the most innovative organizations deploying AI.
• • Your primary responsibilities will involve organizing and guiding our red team to conceptualize and implement novel behaviors and challenges for the Arena. This requires a deep understanding of AI vulnerabilities and creative problem-solving to design engaging and effective tests.
• • You will also supervise the execution of private red-teaming engagements, ensuring that findings are meticulously documented and reported to our clients. This involves managing the entire lifecycle of these engagements, from scoping and planning to execution and final reporting, providing actionable insights to enhance client AI security postures.
• • Effective communication and collaboration are key. You will leverage a suite of tools including Slack, Linear, Discord, our proprietary Arena design editor, and various AI tools to ensure the timely and high-quality delivery of red-teaming results. Your ability to navigate these platforms and integrate them into a cohesive workflow will be critical.
• • Reporting to our VP of Product, you will work closely with our product, marketing, and engineering teams. Furthermore, you will liaise with partners at frontier labs and enterprises, coordinating the efforts of potentially hundreds of red teamers across numerous parallel workstreams. This necessitates exceptional project management skills to align timelines, manage dependencies, and ensure all deliverables are met by contributors you won't directly manage.
• • You will be the voice of our community initiatives, announcing Arena challenges and updates to our growing network of red teamers. A significant part of your role will involve synthesizing product feedback from participants, understanding their experiences, and engaging in in-depth discussions with users via Discord to foster a collaborative and productive environment.
• • To thrive in this role, you should possess significant experience in red teaming or a similar competitive AI challenge environment (like our Arena). You should also have a proven track record in project management, demonstrating your ability to handle complex, multi-stakeholder initiatives.
• • Exceptional writing and organizational skills are paramount. You must be able to articulate complex technical concepts clearly and concisely, manage multiple projects simultaneously, and maintain meticulous records.
• • We are looking for someone who is equally comfortable communicating with highly technical individuals, including red team hackers, AI researchers, and machine learning engineers, as well as with business stakeholders.
• • A strong sense of ownership and initiative is essential. You must be proactive in identifying potential issues, driving solutions, and ensuring that projects are delivered on time and to the highest standard of quality.
• • While direct coding experience is not required, a strong technical product sense is highly valued. You should be able to understand the technical underpinnings of AI systems and security challenges, and contribute to product strategy.
• • We encourage individuals who may not meet 100% of the listed qualifications to apply. Our hiring philosophy prioritizes demonstrated ability and potential over formal credentials.
• • Representative projects you might undertake include overseeing a team of contract red teamers to develop a comprehensive set of indirect prompt injection scenarios for an upcoming Arena challenge, upgrading our private red teaming reporting format to enhance clarity and impact, and triaging reported Arena issues to inform product engineering for feature development and bug fixes. This role offers a unique opportunity to shape the future of AI security and contribute to a critical area of technological advancement.