PDF Annotation & Transcription Experts – Odia

The position of PDF Annotation & Transcription Expert – Odia is offered by Mercor, a company focused on creating training data that improves document‑understanding models for languages and scripts that are currently under‑represented, such as Odia, other Indic scripts, Japanese and Korean. The work revolves around turning real‑world Odia PDFs into richly annotated resources that help AI handle complex layouts and handwritten text. Each day you will locate a publicly available Odia PDF that contains at least one multimodal element—such as an image, table, diagram, or handwritten portion—and note its source. You will then delineate every meaningful region on the page, assign it a component type, and give it a reading‑order index. Relationships between regions and their parent figures or tables must be recorded, and you will transcribe the text exactly as it appears, flagging any illegible sections. Additional metadata about language, document type, dimensions and special features will be captured. After completing a task, a second Odia specialist will review it, and experienced annotators may also perform peer reviews. The role requires native‑level fluency in Odia, including full mastery of the script, its diacritics and conjunct forms. Candidates should have prior experience with document‑related activities such as annotation, transcription, translation, localization, proofreading, journalism or regional‑language data review. Precision at the character level, a systematic approach to applying a taxonomy across many pages, and comfort with unfamiliar layouts like multi‑column newspapers or exam papers are essential. Desirable background includes work on regional‑language AI data, transcription or localization pipelines, OCR correction, or document production tasks such as typesetting and digitization. Familiarity with Unicode normalization, Odia input methods and numeral accuracy is a plus. Success is measured by complete, correctly bounded regions, accurate reading order, flawless character‑by‑character transcription, and passing the second‑expert review on the first attempt.