TextRadar

A closer look
at the work.

At TextRadar I designed the architecture and developed much of the application alongside a multidisciplinary team. The product analysed one or two documents to identify inconsistencies and make them easier to review.

My contribution

• Document segmentation using embeddings and textual structure to identify related passages.
• Comparing passages with language models and locating the text behind each potential inconsistency.
• An interface for reviewing, editing, accepting or rejecting inconsistencies and downloading documents with tracked changes.
• Development of a product for publishers that extracted metadata and generated records and promotional texts from book covers and interiors for review and submission to DILVE.
• Architecture and database design, backend and frontend development, and team coordination.

Profile based on my professional experience. The illustration is a visual interpretation of the field.