AI-in-Healthcare researcher with 5+ years building interpretable deep learning pipelines for clinical digital pathology. Led development of a WSI diagnostic pipeline trained on 2,500+ multi-site cases, reducing diagnostic turnaround from weeks to minutes.
I am a Research Data Analyst in the Division of Computational Pathology at the Indiana University School of Medicine, where I translate AI research into diagnostic and prognostic tools.
My work centers on developing interpretable deep learning models for whole-slide image analysis, from WHO 2021 glioma classification to prognostic risk stratification, bridging the gap between computational research and real-world pathology workflows.
Previously at UPenn, I built multimodal GBM prognostic models achieving AUC around 0.75 across multi-institutional cohort. My research has been published in Neuro-Oncology, Modern Pathology, Nature Scientific Reports, and presented at MICCAI, SNO, ECDP, and AACR.
Led development of interpretable AI models for WSI diagnostic classification & prognostic stratification. Built GPU-backed inference pipeline (A6000/A100 SLURM) enabling 10-min/slide inference vs. multi-week molecular workflows.
Developed multimodal GBM prognostic models (imaging + clinical features), AUC ~ 0.75 across multi-institutional cohorts. Co-authored manuscripts in Modern Pathology and Frontiers in Neuroscience.
QA for cloud migration of transaction modules (legacy → AWS / Java/Spring Boot) for a leading U.S. financial client. Led 6-person testing component across multi-team migration.
Built edge inference pipelines for ANPR and mask-detection.
AI-driven classification of glioma subtypes directly from H&E-stained slides, validated on 2,500+ cases across 25+ sites. Published in Neuro-Oncology. Won Best Paper at CCBB 2025.
Interpretable AI for glioma IDH mutation status prediction directly from WSI, removing dependency on expensive molecular testing. Published in Neuro-Oncology Advances.
View PublicationMultimodal explainable AI integrating imaging + clinical features for glioblastoma patient stratification. AUC ~ 0.75 on multi-institutional test set. Published in Modern Pathology.
View PublicationEnd-to-end clinician-facing pipeline with GPU-backed inference (A100/H100 SLURM) enabling slide analysis in ~10 minutes vs. multi-week molecular workflows. Staged for IU Health clinical validation.
View ProjectsOpen to research collaborations, speaking invitations, and discussions about AI in digital pathology.