Project Spring – Help Build High-Quality Multilingual Healthcare Data

About this role

Use your language expertise to improve how spoken conversations and written summaries align across healthcare content. Language is at the heart of effective communication in healthcare. When medical conversations are transcribed and summarized, critical details such as names, organizations, locations, and other Protected Health Information (PHI) must be captured accurately and consistently. Even small discrepancies can reduce data quality and limit the usefulness of AI systems trained on this information. As a Multilingual PHI Annotation Specialist for Project Spring, you will review conversational audio and corresponding written summaries to ensure that the same PHI entities are correctly identified and aligned in both formats. Your work will help create high-quality multilingual datasets used to improve healthcare-related language technologies. What You’ll Do Your primary task is to annotate and validate PHI entities across spoken and written healthcare content. This includes: Listen to conversational audio recordings Review corresponding written summaries Identify and label PHI entities, including: Person names Organizations Locations Related identifying references Ensure that entities appearing in the audio are accurately reflected in the written summary Detect omissions, inconsistencies, and annotation errors Follow detailed annotation guidelines and quality standards Perform self-review to ensure accuracy and consistency before submission Why This Matters Artificial intelligence is increasingly used to process and summarize healthcare conversations. For these systems to perform reliably, they must accurately recognize and align key information across both spoken and written formats. Your annotations will help build comparable multilingual corpora that improve the performance of language models in healthcare settings, ultimately supporting better transcription, summarization, and information extraction technologies. Available Languages We are currently re