Part Time AI Annotation & Policy Writer

TaskUs is looking for a skilled freelance AI Annotation & Policy Writer to help shape how AI systems learn to understand the world. You will create clear and comprehensive annotation guidelines for our AI data-labelling projects. About the RoleThis is a crucial role where you'll be responsible for producing decision-safe instructions that ensure high-quality, consistent labels across our teams of annotators with minimal oversight. You will independently scope out projects, quickly learn the domain, and write guidelines that are easy to understand and follow. What You'll DoDesign and Author Guidelines: Produce detailed, versioned guidelines that explain the project's purpose, define each label with precision, and provide clear examples and counter-examples. You'll also specify how to handle ambiguity and tricky edge cases.Create Structured Resources: Develop resources like decision trees, visual examples, and a living FAQ to enable quick onboarding and consistent execution. You'll also set up clear escalation paths for any "unknown" issues.Operate Independently and Become the Subject Matter Expert: Self-learn from the brief and sample data, master the taxonomy quickly, deliver clear, decision-safe guidelines with minimal support. Key ResponsibilitiesUnderstand project goals, annotation workflow, and common pitfalls.Translate complex taxonomies into clear, example-rich guidelines.Build visual resources to ensure global teams label data consistently.Draft guidelines that cover:Task purpose and how it impacts the AI model.A precise label taxonomy with examples for in-scope and out-of-scope data.Step-by-step annotation rules, including priorities and tie-breakers.A catalog of edge cases and a policy for handling ambiguity.A library of example annotations with clear and sound explanations.Practical guidance notes and tips for the chosen annotation tool, to ensure remote annotators are set-up for success.Set up a lightweight change-management system, including version history and release notes.Provide recommendations for tool-specific settings on platforms like Labelbox or Dataloop, where relevant.Proactively offer constructive, well-reasoned ideas where you see opportunities to improve the guidelines, annotation workflow, and supporting materials.Incorporate one to two rounds of consolidated written feedback from stakeholders. QualificationsProven Experience: at least 1 year of experience in writing detailed annotation or content-policy guidelines for machine learning projects. Please provide portfolio samples.Exceptional Communication: The ability to simplify complex rules using clear language, visual examples, and decision trees.Collaborative Skills: Hands-on experience working with data scientists or subject matter experts to define annotation taxonomies, schemas, and workflows.Independent and Proactive: Comfortable working with a limited brief, conducting independent research, interviewing stakeholders, and proposing pragmatic solutions when requirements are incomplete.Platform Awareness: Knowledge of one or more data-labeling platforms (e.g., Labelbox, Dataloop).Excellent Writing: Flawless written English and a strong editorial discipline. How to Apply: To apply, please submit your CV and complete the application form.Please include one or two examples of your writing in English, ideally guidelines or structured documentation you’ve authored. Shortlisted candidates may be invited to complete a case study to demonstrate their approach to clarity and structure. Successful candidates will join our TaskVerse community and gain access to additional opportunities in the AI space. Important Note: In this role your work may include exposure to sensitive or potentially challenging content. It is essential that you are comfortable handling such material professionally and objectively, as this work is critical for building safer and more ethical AI systems.