Remote AI Safety & Mental Health Expert (Contract)

mpathic; Seattle (WA); On-site; USD 13,776 - 55,104; Full time

Job summary

mpathic.ai is seeking contract, project-based mental health experts to support AI safety and mental health policy for large language models. You will collaborate on confidential initiatives, roleplaying clinical scenarios with AI agents and providing expert feedback on safety metrics.

Responsibilities include designing AI chat scenarios, conducting qualitative analysis, and developing behavioral taxonomies while upholding strong ethics and confidentiality.

Qualifications

• Graduate-level training in psychology, counseling, social work, psychiatry, or related field.
• Experience with crisis intervention, risk assessment, and severe psychopathology.
• Familiarity with AI tools and LLMs; comfortable with confidential content.
• Ability to work remotely using Slack, LLM tools and Office suites.
• Willingness to sign NDAs and handle sensitive information.

Responsibilities

• Designing and testing AI chat scenarios for safety.
• Conducting qualitative conversation analysis.
• Developing behavioral taxonomies, personas, and evaluation rubrics.
• Providing expert feedback on internal mental health policy.
• Collaborating with engineers and researchers on safety metrics.
• Documenting edge cases and model failures.
• Supporting testing and fine-tuning through structured scenarios.
• Maintaining confidentiality and QA standards.

Skills

Crisis intervention

Ethics and confidentiality

Communication

Collaboration

Remote work

Education

Graduate-level psychology training

Tools

Slack

LLM tools

Microsoft Office

Job description

mpathic.ai is seeking contract, project-based mental health experts to support AI safety and mental health policy for large language models. You will collaborate on confidential initiatives, roleplaying clinical scenarios with AI agents and providing expert feedback on safety metrics.

Responsibilities include designing AI chat scenarios, conducting qualitative analysis, and developing behavioral taxonomies while upholding strong ethics and confidentiality.