Remote Part-Time Financial AI Risk & Safety Specialist

Portland Seed Fund; Seattle (WA); Hybrid; USD 41,000 - 276,000; Part time

Job summary

Mpathic is seeking a part-time Financial Risk & Safety Specialist to evaluate AI system behavior in consumer-facing financial interactions. You will review AI responses and multi-turn conversations to identify risks related to financial guidance, overconfidence, and uncertain disclosures.

This role focuses on adversarial thinking and expert judgment applied to AI outputs in simulated scenarios, with around 10 hours per week for eight weeks and remote collaboration via Slack.

Qualifications

• Minimum professional experience in finance, banking, or risk-related roles.
• Strong ability to assess risk, uncertainty, and appropriateness of financial guidance.
• Interest in red teaming or safety evaluation of AI systems.
• Excellent written communication and clear explanation of reasoning.
• Comfort with remote collaboration using typical productivity tools.

Responsibilities

• Identify risks in AI-generated financial content and conversations.
• Evaluate for unsafe or misleading financial guidance and inappropriate claims.
• Assess handling of uncertainty, disclaimers, and knowledge scope.
• Identify overconfidence, sycophancy, or inappropriate agreement patterns.
• Review multi-turn conversations for drift and policy adherence over time.
• Support red teaming and adversarial testing of AI systems.
• Provide structured feedback and document edge cases and failures.

Skills

Finance

Investment analysis

Financial advising

Banking / wealth mgmt

Regulatory / compliance

Written communication

Tools

Slack

Job description

Mpathic is seeking a part-time Financial Risk & Safety Specialist to evaluate AI system behavior in consumer-facing financial interactions. You will review AI responses and multi-turn conversations to identify risks related to financial guidance, overconfidence, and uncertain disclosures.

This role focuses on adversarial thinking and expert judgment applied to AI outputs in simulated scenarios, with around 10 hours per week for eight weeks and remote collaboration via Slack.