AI Trainer (AI Safety)

TL;DR

AI Trainer (AI Safety): Reviewing and evaluating AI conversations and model outputs with an accent on safety, policy compliance, and accuracy. Focus on identifying harmful behaviors, ranking model responses, and contributing to training datasets for AI safety layers.

Location: Remote (Global — no matter where you work from)

Salary: $30,000 – $50,000 + Equity

Company

An AI Safety company building the safety, reliability, and optimization layer for AI systems through natural-language policies.

What you will do

• Review and evaluate AI conversations, assessing responses for safety, quality, accuracy, and user intent.
• Identify harmful, unsafe, or misleading model behavior and moderate sensitive content.
• Label and categorize model outputs according to internal evaluation frameworks.
• Compare, rank, and score model responses to improve system performance.
• Provide structured feedback to researchers and engineers to refine annotation processes.
• Contribute to datasets used for training and evaluating AI systems.

Requirements

• Exceptional attention to detail and ability to make consistent decisions across large data volumes.
• Strong written English skills with the ability to explain reasoning clearly.
• Ability to follow complex guidelines while exercising good judgment in nuanced situations.
• Curiosity about AI systems and how they function.
• Comfort working with sensitive, offensive, or disturbing content.

Nice to have

• Experience in content moderation, trust & safety, quality assurance, or policy enforcement.
• Experience in data annotation, AI evaluation, RLHF, or model assessment.
• Extensive experience using AI tools and understanding their strengths and limitations.

Culture & Benefits

• Competitive salary of $30k–$50k plus equity.
• Paid time off according to local regulations, regardless of your location.
• Provision of all necessary hardware, tools, and services.
• Opportunity to work in a small, highly focused team on hard AI safety problems.

Hiring process

• Introductory call with HR (25 min).
• Technical take-home assignment.
• Final conversation with the CEO (35 min).