Fluent Language Skills Required: Japanese. Native fluency in Japanese, including full command of kanji, hiragana and katakana, is required for this position. All annotation and transcription work is performed in Japanese.
Why This Role Exists Document understanding breaks down fastest in the languages that parsing and vision-language models rarely see. This project builds training data for exactly those languages: Japanese, alongside Korean and five Indic scripts. Each task takes a real, publicly available PDF page and produces a complete structural map of that page, paired with a faithful transcription of every text region in the original script.
The dataset deliberately concentrates on the material models handle worst: handwriting, dense multi-column layouts, vertical text, tables, diagrams, and mixed-script pages. Documents are drawn from newspapers, textbooks, examinations, and everyday formats such as flyers, forms, manuals, menus, brochures, notices and worksheets, so that the corpus reflects the real diversity of Japanese documents rather than a narrow band of easily parsed ones.
Delivered work is human-authored throughout. Component identification, component typing, reading order and all transcription are performed by people, not generated by parsing models.
What You'll Do Open and check a task: pages are provided, so you do not source documents yourself. We find the PDFs and upload them for you. Before annotating, confirm the page is in Japanese, is legible, has real content, and shows no personal details

Annotate structure: identify and bound every meaningful region of the page - document title, section heading, paragraph, list, table, figure, diagram, caption, formula, question, answer field - and assign each a component type and a reading-order index

Record relationships: link each region to the figure or table it belongs to through a parent component identifier

Transcribe faithfully: reproduce all text exactly as it appears, including kanji, hiragana, katakana, furigana and handwritten content, flagging any region where the source is not legible

Capture page metadata: language, document type, source, page dimensions, and flags for tables, formulas and handwriting

Review a colleague's work: every task is reviewed end to end by a second Japanese expert, and experienced annotators take on that review

Who You Are You are a native Japanese speaker with full command of kanji, hiragana and katakana, including furigana and variant character forms

You have professional experience in interpretation, journalism, transcription, translation, editorial work, or comparable document-intensive work

You are exact: character-level accuracy matters more here than speed, and a single wrong character is a defect

You are systematic: you apply a taxonomy consistently across hundreds of pages rather than improvising per document

You are comfortable with unfamiliar layouts: vertical text, multi-column newspapers, exam papers, handwritten forms

Nice-to-Have Specialties AI training data: annotation, labeling, grading, or bilingual evaluation for training datasets

Transcription and localization: MTPE, subtitling, bilingual QA, OCR correction or post-editing

Document production: typesetting, copy-editing, proofreading, or digitization of Japanese-language material

Script and encoding: Unicode normalization, Japanese input methods, full-width and half-width forms, and kanji variant handling

What Success Looks Like Every meaningful region on the page is captured, correctly bounded and correctly typed

Reading order reflects how the page is actually read, including vertical text and multi-column layouts

Transcriptions match the source character for character, in Japanese script rather than romaji

Your tasks pass second-expert review the first time

Unsuitable pages are flagged up front rather than after thirty minutes of work

Why Join Mercor Build the training data that makes document AI work in scripts it currently handles badly

Work from real published Japanese documents rather than synthetic or templated pages

Quality leads on this project: accuracy is the first measure, with handling time tracked alongside it
•