PDF Annotation & Transcription Experts – Malayalam

Mercor is building a multilingual training set for document‑understanding AI and is looking for PDF Annotation & Transcription Experts who work in Malayalam. The role focuses on creating high‑quality, human‑authored annotations and transcriptions of real‑world Malayalam PDFs, ranging from newspapers and textbooks to flyers and forms, so that future models can handle the scripts they currently struggle with. Each day you will receive a PDF page that has already been collected and uploaded. Your tasks are to verify that the page is in Malayalam, legible and free of personal data; to mark every meaningful region (titles, headings, paragraphs, lists, tables, figures, diagrams, captions, formulas, question‑answer fields) with precise bounding boxes, component types and reading‑order indices; to link each region to its parent element; to transcribe the text exactly as it appears, including handwritten material; to record metadata such as language, document type, source and page dimensions; and finally to review a fellow expert's completed task to ensure it meets the required standards. The position requires native‑level fluency in Malayalam, including full mastery of the script, its diacritics and conjunct forms. You must have prior experience handling documents in roles such as annotation, transcription, translation, localization, proofreading, journalism or regional‑language data review. Accuracy at the character level is essential, as a single misplaced diacritic is considered a defect. You should apply the provided taxonomy consistently across hundreds of pages and feel comfortable with diverse layouts, including multi‑column newspapers, exam papers and handwritten forms. While not mandatory, the team values additional expertise that can accelerate the project. Experience with regional‑language AI data annotation, labeling or bilingual evaluation is a plus. Background in transcription‑related work such as subtitling, OCR correction or post‑editing is also welcomed. Familiarity with document production tasks, typesetting, copy‑editing, proofreading, digitisation of Malayalam material, and with Unicode normalization or Malayalam input methods further strengthens your profile. The project emphasizes quality above speed, and successful contributors see their work pass a second‑expert review on the first attempt. What the role asks for: - Native fluency in Malayalam (must‑have) - Experience with document work (annotation, transcription, translation, localization, proofreading, journalism, or regional‑language data review) (must‑have) - Ability to achieve character‑level accuracy (must‑have) - Systematic application of taxonomy across many pages (must‑have) - Comfort with multi‑column, handwritten, and mixed‑script layouts (must‑have) - Experience in regional‑language AI data annotation or evaluation (nice‑to‑have) - Experience in transcription, subtitling, OCR correction or post‑editing (nice‑to‑have) - Experience in document production (typesetting, copy‑editing, proofreading, digitisation) and Unicode handling (nice‑to‑have)