Designed and developed a multimodal deep learning classification system on MRI medical images for Alzheimer's disease progression prediction. Evaluated and adopted FastSurfer over FreeSurfer for brain segmentation, reducing per-subject preprocessing from hours to ~1 minute and enabling large-scale iterative experiments across 2,000+ subjects (10,000+ scans). Architected an experiment management platform (FastAPI + SSE + SQLite) supporting YAML-driven hyperparameter combinations, real-time training monitoring, Grad-CAM interpretability analysis, and NiiVue medical image viewing. Developed a digital care platform for neurodegenerative disease management — patient-facing cognitive tracking app (React), clinician dashboard (React), and FastAPI backend with SQLite. Integrated a RAG-based chatbot (LangChain) grounded in health education materials for caregiving guidance. Collaborated directly with clinicians to translate requirements into system specifications.
OPERATIVE_PROFILE
LOADING_CREDENTIALS...
DION
Graduate training centered on multi-omics analysis, data integration, and biomarker discovery using machine-learning approaches. Developed end-to-end analytical workflows—from pipeline construction to downstream processing, visualization, and result interpretation.
Built an NGS/qPCR gut microbiome analysis pipeline from scratch using Python, QIIME2, and pandas, deployed on GCP. Handled 16S amplicon preprocessing, taxonomy refinement via BLAST + NCBI, and automated report generation. First engineering role — transitioned from wet-lab biology to independent data pipeline development.
Studied organic chemistry, biochemistry, molecular biology, and standard wet-lab experimental techniques. Joined a structural bioinformatics laboratory, where I began working with protein structure analysis and gained my first experience in computational biology.