About this role
Role Overview
Help create high-quality training data that improves document AI for Telugu and other scripts that are often poorly handled by parsing and vision-language models. You will work with real, publicly available Telugu PDF pages, building a complete structural map of each page and transcribing every text region accurately in Telugu script. The work focuses on challenging material, including handwriting, dense multi-column layouts, tables, diagrams, and mixed-script pages.
Documents may include newspapers, textbooks, examinations, flyers, forms, manuals, menus, brochures, notices, and worksheets. All annotation and transcription work is performed in Telugu and is completed by human experts rather than parsing models.
Key Responsibilities
- Find publicly accessible Telugu PDFs in assigned document types that contain at least one multimodal element, such as an image, table, diagram, or handwriting, and record the source.
- Identify, bound, and classify every meaningful page region, including titles, headings, paragraphs, lists, tables, figures, diagrams, captions, formulas, questions, and answer fields.
- Assign a reading-order index to each component and link related regions to their parent figure or table using a parent component identifier.
- Transcribe text exactly as it appears in Telugu script, including handwritten content, and flag regions that are not legible.
- Capture page metadata, including language, document type, source, page dimensions, and indicators for tables, formulas, and handwriting.
- Review a colleague''s work end to end when assigned as an experienced annotator.
Qualifications
- Native Telugu fluency, with full command of Telugu script, orthography, diacritics, and conjunct forms.
- Experience with document-focused work such as annotation, transcription, translation, localization, subtitling, proofreading, journalism, or regional-language data review.
- Strong character-level accuracy, including careful handling of diacritics.
- A systematic approach to applying a taxonomy consistently across many pages.
- Comfort working with unfamiliar layouts, including multi-column newspapers, examination papers, and handwritten forms.
Preferred Experience
- Regional-language AI data work, including annotation, labeling, grading, or bilingual evaluation for training datasets.
- Transcription or localization work, such as MTPE, subtitling, bilingual QA, OCR correction, or post-editing.
- Telugu document production, including typesetting, copy-editing, proofreading, or digitization.
- Knowledge of Unicode normalization, Telugu input methods, and accurate numeral and character forms.
What Success Looks Like
- Every meaningful page region is captured, correctly bounded, and correctly typed.
- Reading order matches how the page is read, including across columns.
- Transcriptions match the source character for character in Telugu script, without transliteration.
- Work passes second-expert review on the first submission.
- Newly sourced documents add meaningful layout diversity rather than repeating existing templates.
Work Terms
- Remote hourly engagement.
- Accuracy is the primary quality measure, with handling time also tracked.
Compensation
- $12.68 per hour.