Malayalam Document Annotation & Transcription Pro

Mercor; Brussel Hoofdstad; On-site; EUR 30,000 - 46,000; Full time

Job summary

Mercor is seeking native Malayalam speakers for a data annotation project to build training data for multilingual document AI. You will annotate structure, bound regions, and transcribe Malayalam text with high accuracy, including handwritten content.

Work involves finding publicly accessible Malayalam PDFs, identifying page regions such as titles, headings, paragraphs, tables and figures, and ensuring reading order across columns, with two-pass review by a Malayalam expert.

Qualifications

• You are a native Malayalam speaker with full command of the script and its diacritics.
• Experience with document annotation, transcription, localization, subtitling, proofreading, or data review in a language context is helpful.
• Character-level accuracy matters; correct diacritic and conjunct forms are essential.

Responsibilities

• Source documents: locate publicly accessible Malayalam PDFs with multimodal elements (images, tables, diagrams, handwriting).
• Annotate structure: identify and bound every meaningful region (title, heading, paragraph, list, table, figure, caption, formula) and assign type and reading order.
• Record relationships: link each region to its parent figure or table via a component identifier.
• Transcribe faithfully: reproduce all text exactly as it appears in Malayalam, including handwritten content.
• 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 Malayalam expert.

Job description

Mercor is seeking native Malayalam speakers for a data annotation project to build training data for multilingual document AI. You will annotate structure, bound regions, and transcribe Malayalam text with high accuracy, including handwritten content.

Work involves finding publicly accessible Malayalam PDFs, identifying page regions such as titles, headings, paragraphs, tables and figures, and ensuring reading order across columns, with two-pass review by a Malayalam expert.