At Biologix, I designed a machine learning model to detect atrial fibrillation from wearable-acquired PPG signals. It now screens over 10,000 exams per month, enabling early cardiovascular risk detection.
I also calculated heart rate variability metrics across time, frequency, and geometric domains for a large patient cohort. I formulated hypotheses, conducted statistical validation, and co-authored research articles and medical conference posters.
At Centro Pi, IMPA, I developed and optimized OCR algorithms to process up to 20 million handwritten documents for OBMEP. I also created Python-based machine learning pipelines to develop and assess large multimodal Vision-Language deep learning models.
