AI Guidelines Writer – Linguistics / Instructional Design
Pay: $45–$65/hour
Location: United States
Schedule: 35+ hours/week, weekdays
Work Setup: Remote
About the Role
This opportunity is for experienced Linguists, Instructional Designers, and Technical Writers to help create clear and consistent guidelines used in AI training and evaluation.
You’ll take complex or ambiguous requirements and turn them into practical instructions and rating rubrics that human evaluators can apply consistently.
Projects may cover a variety of subject areas, including finance, retail, insurance, legal, and sports.
What You’ll Do
- Turn complex project requirements into clear evaluator instructions.
- Write and refine rating guidelines, rubrics, and evaluation criteria.
- Identify ambiguity, contradictions, and missing information in existing guidelines.
- Develop guidance for difficult or unusual edge cases.
- Translate subject-specific requirements into easy-to-follow instructions.
- Collaborate with subject-matter experts and project teams.
- Improve guidelines based on feedback and recurring evaluator questions.
Qualifications
- 3+ years of experience in linguistics, instructional design, technical writing, or a related field.
- Experience creating or refining guidelines and rubrics for human evaluators.
- Previous exposure to GenAI, RLHF, AI evaluation, or AI training-data projects.
- Excellent written communication and editing skills.
- Strong ability to identify and resolve ambiguity in written instructions.
- Comfortable working across different subject areas.
- Able to provide examples of improving unclear or contradictory guidelines.
- Available for at least 35 hours per week on weekdays.
Ideal Candidate
You’re skilled at taking complicated requirements and making them simple, precise, and difficult to misinterpret. You have a strong eye for inconsistencies and can anticipate situations where evaluators may interpret the same instruction differently.
Experience with AI evaluation, RLHF, annotation guidelines, rubric design, or human-feedback programs is particularly valuable.