Founder & Lead Developer—MedScribe AI (AI model evaluation, clinical scenario prompt engineering, supervised annotation)
As Founder & Lead Developer of MedScribe AI, I designed and evaluated AI models for structured clinical documentation in Nigerian teaching hospitals. My work included prompt engineering, stress-testing of clinical prompts, and annotating AI outputs for accuracy and logical consistency. I implemented and executed supervised training workflows for multimodal clinical data, including text, audio, and images. • Conducted multimodal data quality evaluation of clinical notes, including both AI-generated and human-generated content. • Annotated clinical decision support outputs against verified medical guidelines and identified model deviations. • Iteratively generated prompts and assessed model responses for hallucinations, factual errors, and logical inconsistencies. • Developed role-based annotation workflows emulating clinical review hierarchies and feedback loops.