Developing CNNs (ResNet, DenseNet) and vision architectures for lesion classification, pneumonia triage, and Grad-CAM interpretability.
Designing machine learning pipelines (XGBoost, survival modeling) for clinical risk stratification and intensive care triage using MIMIC-IV.
Ensuring model features align with clinical pathological reality, minimizing confounding factors and maximizing actionable diagnostic decision support.
Deep learning implementations, clinical decision support architectures, and EHR predictive pipelines.
Deep learning framework using transfer learning on ResNet-34 for automated pediatric chest X-ray diagnosis, distinguishing viral vs. bacterial etiology with Grad-CAM heatmaps.
Machine learning predictive pipeline trained on the MIMIC-IV EHR database to forecast postoperative complications, ICU length of stay, and 30-day mortality using SHAP explanations.
2D/3D U-Net deep convolutional architecture with attention gates and compound Dice + Focal loss for precise semantic segmentation of pathological tumor margins on diagnostic scans.
A unique blend of rigorous medical education, high-volume clinical rotations, and computational specialization.
I am always eager to discuss computational medicine, medical AI research, PhD openings, or collaborative deep learning projects.