Japanese-Speaking Multimodal Image Evaluation Annotator

You will create realistic evaluation materials for AI systems that work with images. The work combines image sourcing, web research, copyright and provenance checks, image editing, prompt writing, and clear documentation.

• Source high-quality photographs connected to Japan, including landmarks, universities, municipal buildings, products, and domestic film or media.
• Confirm whether each image is original, public domain, copyright-free, or explicitly licensed, and record its provenance and license type.
• Use Google Search, Google Lens, reverse image search, and reliable supporting sources to identify subjects from partial visual clues.
• Crop, occlude, mask, or degrade images to hide identifying details while keeping the remaining visual information realistic.
• Write time-bound prompts in English, one-sentence grading criteria, and solution notes covering the true identity, research process, evidence, and expected result.

What it pays and takes

This is an entry-level remote contractor role. A bachelor's degree or equivalent practical experience is suitable, and previous AI evaluation, data annotation, content review, quality assurance, or analytical experience is helpful but not required.

• Pay: Not specified in this listing.
• Schedule: At least 20 hours per week, with four hours of overlap with Pacific Time.
• Engagement: Part-time contract work for up to 16 weeks.
• Location: Remote and open worldwide.
• Languages: Highly fluent Japanese and strong working proficiency in English.
• Knowledge: Strong familiarity with culturally significant but non-touristy Japanese places, institutions, products, and media.
• Research: Ability to investigate iteratively and verify conclusions from partial visual clues.
• Image rights: Understanding of licensing, copyright, provenance, public-domain sources, and Creative Commons sources.
• Tools and writing: Basic ability to crop, mask, or annotate images, plus careful and objective writing for prompts, rubrics, and research notes.
• Work process: Use shared folders, file-naming conventions, and structured deliverable submissions.

How it works

Apply on OpenTrain with your resume and then complete the application on the hiring site.

About AI training work

AI training is the human work behind systems that understand images, text, audio, and other data. Contributors prepare examples and judge results so AI models can improve, and people with strong language, research, or subject knowledge can bring useful expertise to this process.