Cybersecurity for AI Safety (Contract)

mpathic; United States; Remote; USD 83,000 - 165,000; Full time

Job summary

mpathic is seeking cybersecurity experts to help train and evaluate next-generation AI systems. You will design realistic cybersecurity prompts and assess AI-generated responses for technical accuracy, ensuring models handle sensitive topics responsibly.

You will work with a multidisciplinary team to improve AI understanding of software security, cyber defense, vulnerability research, and related domains, shaping the quality and reliability of AI systems used by millions.

Qualifications

• Bachelor's or Master's degree in Computer Science, Cybersecurity, or a closely related technical field.
• 5+ years of professional software engineering experience at a high-growth startup or established technology company.
• Deep knowledge of cybersecurity principles, modern software systems, and secure development practices.

Responsibilities

• Design realistic, expert-level cybersecurity prompts spanning areas such as application security, network security, cloud infrastructure, vulnerability management, and secure software development.
• Evaluate AI-generated responses for technical accuracy, completeness, and practical usefulness.
• Assess how models handle sensitive and dual-use cybersecurity scenarios, ensuring responses are responsible and aligned with project guidelines.
• Apply structured evaluation rubrics to annotate prompts and model outputs with consistency and precision.
• Provide clear, actionable feedback that helps improve the performance, safety, and reliability of frontier AI systems.

Skills

Cybersecurity

Software engineering

Technical reasoning

Written English communication

Education

Bachelor’s or Master’s in Computer Science or Cybersecurity

Tools

Red-teaming

Job description

About mpathic.ai

mpathic is keeping humans safe in the AI era through automated tools and expert datasets that are rooted in psychology and powered by clinicians. We are a series A start-up backed by Tier 1 investors including Foundry.vc and Next Frontier Capital.

If you have applied for an expert role within the past 12 months, you do not need to reapply.

About the role

We're looking for cybersecurity experts to help train and evaluate the next generation of AI systems. In this role, you'll apply your technical expertise to create realistic cybersecurity scenarios, assess AI-generated responses, and help ensure models are accurate, useful, and responsible when handling complex security topics.

You’ll work alongside a multidisciplinary team improving how AI understands software security, cyber defense, vulnerability research, and related technical domains. Your expertise will directly shape the quality, safety, and reliability of AI systems used by millions.

What you'll do

• Design realistic, expert-level cybersecurity prompts spanning areas such as application security, network security, cloud infrastructure, vulnerability management, and secure software development.
• Evaluate AI-generated responses for technical accuracy, completeness, and practical usefulness.
• Assess how models handle sensitive and dual-use cybersecurity scenarios, ensuring responses are responsible and aligned with project guidelines.
• Apply structured evaluation rubrics to annotate prompts and model outputs with consistency and precision.
• Provide clear, actionable feedback that helps improve the performance, safety, and reliability of frontier AI systems.

Qualifications

You meet one of the following qualifications:

• Bachelor's or Master's degree in Computer Science, Cybersecurity, or a closely related technical field.
• 5+ years of professional software engineering experience at a high-growth startup or established technology company.

In addition, you have:

• Deep knowledge of cybersecurity principles, modern software systems, and secure development practices.
• Excellent technical reasoning and strong written communication skills in English.
• Sound judgment when working with sensitive or dual-use security topics.
• The ability to apply detailed guidelines while maintaining high-quality, consistent evaluations.

Nice to have

• Experience in offensive security, penetration testing, vulnerability research, security engineering, or a related discipline.
• Previous experience evaluating, reviewing, benchmarking, or red-teaming technical or AI-generated content.
• Familiarity with AI evaluation, model assessment, or human-in-the-loop data annotation workflows.

Compensation

This is a contract, project-based position. The hourly rate will depend on experience and project parameters. The rate will be disclosed prior to starting the project.