PDF Annotation & Transcription Experts – Bengali

The role of a PDF Annotation & Transcription Expert in Bengali involves working on a project that aims to create training data for document understanding in languages that are less commonly handled by parsing and vision-language models. This includes Bengali, alongside other Indic scripts, Japanese, and Korean. The focus is on producing a complete structural map of real, publicly available PDF pages, paired with a faithful transcription of every text region in the original script. The documents cover a wide range of formats, from newspapers and textbooks to everyday items like flyers and forms, ensuring the dataset reflects the true diversity of Bengali documents. The main day-to-day responsibilities include opening and checking tasks, annotating the structure of each page, recording relationships between different regions, transcribing text faithfully, and capturing page metadata. Each task involves confirming the page is in Bengali, legible, and contains real content. The expert must identify and bound every meaningful region of the page, assign component types and reading-order indices, and link regions to their parent components. Transcription must be exact, reproducing all text as it appears, including handwritten content. Additionally, experts review a colleague's work to ensure accuracy. To excel in this role, candidates must have native fluency in Bengali, including a full command of the script, its diacritics, and conjunct forms. Experience with document work, such as annotation, transcription, translation, or proofreading, is essential. The role requires a high level of accuracy and a systematic approach to applying a taxonomy consistently across hundreds of pages. Comfort with unfamiliar layouts, such as multi-column newspapers and handwritten forms, is also crucial. Nice-to-have skills include experience with regional-language AI data, transcription and localization, and document production. Familiarity with script and encoding, such as Unicode normalization and Bengali input methods, is also beneficial. The team at Mercor values accuracy and quality, and the project offers the opportunity to work on real, published Bengali documents rather than synthetic or templated pages. What the role asks for: - Native fluency in Bengali, including full command of the script, its diacritics, and conjunct forms. - Experience with document work, such as annotation, transcription, translation, or proofreading. - Comfort with unfamiliar layouts, such as multi-column newspapers and handwritten forms. - Experience with regional-language AI data, transcription and localization, and document production (nice-to-have).