At Seguros SURA, I built a clinical database from electronic health records and developed NLP pipelines to extract, clean, and anonymize unstructured clinical text. I also developed and validated a supervised model for early identification of patients at risk of Orphan Disease, using SHAP to interpret its predictions.
At the Faculty of Engineering, Universidad de Antioquia, I developed a student chatbot with a RAG architecture, a FastAPI backend, and LLM APIs. It answered 40% of student requests, complaints, and claims.
At Universidad de Antioquia, I analyzed medical data and researched machine learning for telemedicine. At Mackenzie Presbyterian University, I prepared image data and trained U-Net models for semantic segmentation.
