Malayalam Document Annotation & Transcription Specialist

The role focuses on creating high-quality training data for document understanding models, particularly for underrepresented languages like Malayalam. This project involves annotating and transcribing real, publicly available PDF pages—including handwritten content, dense multi-column layouts, tables, diagrams, and mixed-script pages—to build a robust dataset for parsing and vision-language models.

The dataset is deliberately designed to address the most challenging aspects of document parsing: handwriting, complex layouts, tables, and mixed-script pages. Documents span a wide range of sources, such as newspapers, textbooks, examinations, flyers, forms, manuals, menus, brochures, notices, and worksheets, ensuring the corpus reflects real-world diversity rather than simplified examples.

All work is human-authored, ensuring accuracy in component identification, typing, reading order, and transcription. No generated content is used.

Requirements

What You'll Do

• Open and verify tasks: Confirm that each provided PDF page is in Malayalam, legible, contains real content, and does not include personal details before proceeding with annotation.
• Annotate page structure: Identify and bound every meaningful region (e.g., document title, section heading, paragraph, list, table, figure, diagram, caption, formula, question, or answer field) and assign each a component type and reading-order index.
• Record relationships: Link regions to their parent components (e.g., figures or tables) using parent component identifiers.
• Transcribe faithfully: Reproduce all text exactly as it appears in Malayalam script, including handwritten content. Flag any region where the source is illegible.
• Capture page metadata: Document language, type, source, page dimensions, and flags for tables, formulas, and handwriting.
• Review colleague’s work: Every task is reviewed end-to-end by a second Malayalam expert. Experienced annotators also perform reviews.

Who You Are

• A native Malayalam speaker with full command of the script, including diacritics and conjunct forms.
• Experienced with document-related tasks such as annotation, transcription, translation, localization, subtitling, proofreading, journalism, or regional-language data review.
• Detail-oriented: Character-level accuracy is critical, and even a single incorrect diacritic is considered a defect.
• Systematic: Apply a consistent taxonomy across hundreds of pages rather than improvising per document.
• Adaptable: Comfortable working with unfamiliar layouts, such as multi-column newspapers or handwritten exam papers.

Nice-to-Have Specialties

• Experience with regional-language AI data tasks, including annotation, labeling, grading, or bilingual evaluation for training datasets.
• Transcription and localization expertise, such as MTPE (Machine Translation Post-Editing), subtitling, bilingual QA, OCR correction, or post-editing.
• Document production skills, including typesetting, copy-editing, proofreading, or digitization of Malayalam-language material.
• Familiarity with script and encoding, such as Unicode normalization, Malayalam input methods, numeral-form and character-form accuracy.

What Success Looks Like

• Every meaningful region on the page is accurately captured, bounded, and typed.
• Reading order reflects how the page is actually read, including across columns.
• Transcriptions match the source character-for-character in Malayalam script rather than transliteration.
• Tasks pass second-expert review on the first attempt.
• Unsuitable pages are flagged immediately rather than after spending time on them.