Mindrift collaborates with technology companies to improve AI systems through human-guided evaluation. As a Freelance English Writer - AI Trainer, you will create realistic, structured conversations to test and improve AI performance, focusing on instruction-following, emotional intelligence, and consistency under changing conditions.
Responsibilities:
- Craft multi-turn conversations simulating diverse user personas.
- Stress-test AI models’ responses using a range of tones, registers, and emotional contexts.
- React and adapt to model outputs while maintaining narrative coherence and task focus.
- Identify failure modes and patterns in LLM behavior.
- Document observations clearly and consistently to inform AI improvements.
- Apply structured guidelines, rubrics, or annotation frameworks to assess outputs.
Requirements:
- Undergraduate or postgraduate qualification in Arts, English, Creative Writing, Journalism, MFL, Psychology, Cognitive Science, or equivalent professional experience.
- 1+ years’ experience in Conversational AI Testing, Narrative Design, or Adversarial Model Testing is advantageous.
- C2-level English proficiency (CPE, TOEFL 114+, IELTS 8.0+).
- Strong storytelling and analytical skills to simulate realistic human dialogue.
- Ability to follow structured guidelines and provide clear documentation.
Benefits:
- Fully remote, part-time, project-based work.
- Paid contributions with hourly or fixed project rates, including potential incentives.
- Training on guidelines and structured conversation creation.
- Mentorship support for onboarding and task improvement.
This project-based role provides flexible engagement while contributing to the development of advanced AI conversational systems.