AI Model Policy Trainer, Mental Health (Seattle)

Role Overview

The AI Model Policy Trainer, Mental Health position is a full-time, onsite role based in Seattle, Washington, supporting Handshake AI through TCWGlobal. This opportunity suits professionals with a background in mental health or human services who want to help evaluate and improve how conversational artificial intelligence handles sensitive emotional topics.

Key Responsibilities

Trainers evaluate how conversational AI systems respond to user discussions involving emotional distress, mental-health conditions, crisis situations, unusual beliefs, and other sensitive experiences. Daily tasks include reviewing user conversations and model behavior against detailed safety policies to determine if responses are empathetic, properly grounded, and sensitive to risk without reinforcing harmful beliefs. Professionals in this role apply structured policy taxonomies to classify user requests, examine individual messages as well as longer dialogues, and assess actions taken by AI tools like searches or messaging. Additional responsibilities include writing clear written rationales connected to evidence, participating in team calibration exercises, identifying policy edge cases, and collaborating with researchers and operations teams.

Requirements

Candidates must possess direct experience supporting individuals or families through mental-health or behavioral-health challenges. This background may stem from clinical care, social work, nursing, crisis support, professional caregiving, certified peer support, family advocacy, or care navigation. While clinical licensure is welcomed, it is not mandatory. Applicants must be comfortable working onsite in Seattle from Monday through Friday, 8 a.m. to 5 p.m. PT, and possess strong written communication skills, sound judgment, professional boundaries, and the ability to interpret sensitive conversations with empathy and cultural awareness. Because of the nature of the work, candidates must be able to view potentially distressing material—such as psychosis, self-harm, or severe emotional distress—professionally and sustainably.

Nice to Have

Familiarity with conversational AI tools and prior exposure to structured case reviews, documentation, quality assurance, or rubric-based evaluation are considered a strong plus. Extra credit is given for background working with individuals experiencing crisis, psychosis, paranoia, or suicidal ideation, or experience acting as a certified peer specialist, case manager, or behavioral-health professional. Prior knowledge of popular AI systems like ChatGPT, Claude, or Gemini is also helpful.

Skills and Tools

Salary and Benefits

Compensation for this W-2 assignment ranges from $45 to $65 per hour, with placement within the tier depending on experience. The role is full-time at 40 hours per week, non-exempt, and eligible for overtime pay. Benefits are administered through TCWGlobal and include medical, dental, and vision insurance, flexible spending accounts, pretax commuter benefits, and a 401(k) retirement plan with an employer match. Eligible workers also receive voluntary life and accident insurance, wellness reimbursements, pet insurance discounts, and charitable matching. Paid time off accrues at one hour per 40 hours worked starting on the first day, alongside paid sick leave and 13 paid holidays.

About the Company

Handshake powers a massive career network connecting millions of job seekers, employers, and educational institutions with the shared belief that everyone deserves a path to a great career. Handshake AI collaborates directly with frontier AI lab researchers to publish benchmarks, create evaluations, and improve artificial intelligence models through human expertise. This specific assignment is serviced through TCWGlobal.

How to Apply

To be considered for this position, please submit your application directly on the official company careers page.