Medical AI Data Annotation

I’m building a generative-AI system for the medical domain and have a growing collection of model-generated replies that now need rigorous evaluation. Your job is to read each response and annotate it for clinical accuracy and safety so the data can be fed straight back into the next training cycle. The data: text-based replies produced by my model. What I need from you: a high-accuracy judgment on every reply—accurate, partially accurate, or inaccurate—plus a short corrective note whenever you flag an error or risky advice. The finished labels must arrive in a clean JSON or CSV file that slots seamlessly into our training pipeline. Because this set will drive AI model training, I can only work with someone who has demonstrable clinical knowledge (e.g., practicing clinician, certified medical coder, or researcher with peer-reviewed publications) and previous experience on annotation platforms such as Prodigy, Labelbox, or similar. Acceptance criteria • Every reply reviewed and labeled following the rubric we agree on. • ≥ 99 % inter-rater agreement on a 200-item spot check. • Delivery in the specified schema, zero formatting errors. If you can guarantee that level of detail and accuracy, I’d like to hear how quickly you can turn around an initial batch of 5 000 replies and what tools you prefer to work with.