Japanese Document Transcription & Annotation Specialist

Obsidian; San Francisco (CA); Remote; USD 27,552,000 - 48,216,000; Full time

Job summary

Obsidian is seeking meticulous Japanese annotators to build training data for complex scripts. You will verify page language, ensure legibility, and annotate structure precisely, including bound regions and reading order. Tasks require reproducing text exactly as shown, with attention to kanji, kana, and any handwritten content.

The role emphasizes consistent taxonomy, layout familiarity, and thorough internal reviews to maintain high-quality corpora for model training.

Qualifications

• You are a native Japanese speaker with full command of kanji, hiragana and katakana.
• 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.
• You are systematic: you apply a taxonomy consistently across hundreds of pages.
• You are comfortable with unfamiliar layouts: vertical text, multi‑column pages, handwritten forms.

Responsibilities

• Open and check a task: pages are provided; confirm the page is Japanese, legible, content-rich, and contains no personal details.
• Annotate structure: bound every meaningful region and assign a type and reading-order index.
• Record relationships: link regions to their parent figure or table.
• Transcribe faithfully: reproduce text exactly as it appears, including furigana and handwritten content.
• Capture page metadata: language, document type, source, and page dimensions.
• Review a colleague's work: every task is reviewed end-to-end by a second expert.

Skills

Japanese (native)

Attention to detail

Document annotation

Job description

Obsidian is seeking meticulous Japanese annotators to build training data for complex scripts. You will verify page language, ensure legibility, and annotate structure precisely, including bound regions and reading order. Tasks require reproducing text exactly as shown, with attention to kanji, kana, and any handwritten content.

The role emphasizes consistent taxonomy, layout familiarity, and thorough internal reviews to maintain high-quality corpora for model training.