work

zhongzheng fu lab, ut southwestern (utsw) · dallas, tx

analyzed human intracranial eeg during a stroop color–word task in epilepsy patients with depth macro-electrodes across amygdala, hippocampus, mid-cingulate cortex, orbitofrontal cortex, and pre-sma. preprocessed neuralynx recordings from 3 patients across 25 sessions and built single-trial decoding pipelines with leave-one-subject-out cross-validation, benchmarking linear, tree-based, and deep learning models on spectral features. demonstrated single-trial decodability of stroop conflict and errors (auroc 0.81 and 0.77), error prediction ~630 ms pre-response, and reaction time from neural activity (spearman ρ 0.83); mapped region- and frequency-specific coding with delta/theta conflict signals in mcc and reorganized mcc–pre-sma theta coupling under conflict.

mantis ai, mit csail / kellis lab · cambridge, ma

contributed to ai pipelines for drug discovery and protein–protein interaction prediction. implemented and optimized model components via bug fixes, pull requests, and performance improvements. supported integration testing and documentation for production-ready model releases.

azad bioinformatics lab, university of north texas · denton, tx

developed ml pipelines for pathogen prediction using transformer-based protein language models. designed scalable proteome embedding frameworks for antibiotic resistance inference. integrated biological domain knowledge with statistical inference to reduce feature sparsity in genomic datasets.

diamond bioinformatics · denton, tx

co-founded an independent computational biology initiative spanning biomedical engineering, ai, and neuroscience. led development of a patent-pending ml framework for parkinson's disease early detection. coordinated multi-institutional collaborations and aligned workflow integration across teams.