Medical Imaging Researcher · Université de Rennes / CLCC Eugène Marquis 🇫🇷
Electronic Engineer · Universidad de Antioquia 🇨🇴

I develop machine learning models for clinical decision-making in radiotherapy, with a focus on predicting treatment outcomes from medical imaging. My work sits at the intersection of Gaussian Processes, deep learning, and computer vision applied to real-world healthcare problems.

Currently pursuing graduate studies in the SiVOS program at Rennes, working on PET/CT-based prediction of radiation-induced toxicities in head and neck cancer patients.

Xerostomia prediction from PET/CT imaging — Building predictive models that use salivary gland imaging biomarkers to identify patients at risk of severe dry mouth after radiotherapy. Presented at ESTRO 2026, showing that including ipsilateral gland data significantly improves prediction accuracy.

Gaussian Processes for medical applications — Exploring Bayesian non-parametric approaches (GP regression, Bayesian Neural Networks) for uncertainty-aware clinical predictions.

Human motion analysis & embedded ML — Prior work at Universidad de Antioquia on real-time inference for motion tracking on resource-constrained devices.

📄 Google Scholar · See my publications and citation metrics

ML / DL: PyTorch · TensorFlow · Keras · Scikit-learn · Gaussian Processes (GPyTorch)
Medical Imaging: SimpleITK · 3D Slicer · OpenCV · DICOM pipelines
Infrastructure: Docker · AWS · MLOps · Linux · Git
Languages: Python · C · C++ · Bash

• ESTRO 2026 — Poster presentation on PET/CT-based xerostomia prediction in head and neck cancer
• "Más allá del aula: un viaje hacia la investigación" — Talk at Universidad de Antioquia, inspiring undergraduate students to pursue research careers
• Active member of the ML & Robotics Semillero (UdeA) and IEEE volunteer