Information Security Expert for AI Model Training

Use senior information security expertise to design and execute adversarial tests and red-team exercises against enterprise SaaS and cloud environments, and provide high-quality domain input that helps train next-generation AI systems. This role focuses on real-world security evaluation, threat modeling, and reviewer-level feedback rather than prior AI experience. Your findings will shape how models learn, reason, and respond to adversarial behavior.

• Design and define comprehensive red-teaming workflows and adversarial scenarios targeting SaaS and cloud-based environments.
• Conduct hands-on penetration testing and adversarial threat modeling across platforms such as Microsoft 365, Google Workspace, Slack, GitHub, and Snowflake.
• Provide domain expertise used to train next-generation AI systems by creating realistic, high-quality adversarial inputs and scenarios.
• Act as a senior reviewer, critically evaluating the quality, depth, and realism of red-teaming exercises and deliverables.
• Identify and document potential vectors for misuse or manipulation of AI agents, including prompt injection, privilege escalation, data exfiltration, and destructive actions.
• Contribute practical recommendations for security controls, policies, and detection strategies applicable to real-world SaaS deployments.
• Collaborate with other cybersecurity professionals, giving detailed written and verbal feedback on findings and remediation approaches.
• Support benchmarking efforts by producing adversarial scenarios that reflect contemporary threats and attack methodologies.

• Required skills: cybersecurity, pentesting, red-teaming.
• Extensive hands-on experience in one or more security domains, such as red teaming, penetration testing, application security, cloud security, or security engineering.
• Deep understanding of enterprise SaaS security concepts, including identity management, role based access control and permissions, and data governance.
• Direct familiarity with Microsoft 365, Google Workspace, Slack, GitHub, and Snowflake from a security evaluation perspective.
• Proven ability to think adversarially about how AI and SaaS platforms could be exploited or compromised.
• Strong written and verbal communication skills for clear articulation of complex security issues and recommended remediations.
• Demonstrated track record of practical, real-world impact on security projects, beyond descriptive resume experience.
• No prior experience in AI is required; domain expertise is the primary qualification.

• Prior experience in AI or large language model security, or adversarial testing of agent-based systems.

• Role type: Contractor.
• Location: Remote.
• Engagement: Contribute to a high-impact project for an enterprise SaaS customer, collaborating with other security professionals and reviewers.

Hourly rate between $60 and $100 per hour.

micro1 is an AI data lab that converts domain expertise into training data, evaluations, and feedback loops to improve frontier AI systems. Experts from diverse fields provide real-world input used to train and evaluate models, enabling higher quality model behavior.