cv

Research Scientist

Meta FAIR

• Worked on developing foundation models for EEG (more info soon!).

Medical Informatics PhD Student

University of California, Los Angeles

• Developing a foundation model for EEG using deep state-space models and graph neural networks with over 10,000 hours of EEG data.
• Using a novel clinical dataset to better understand neural connectivity in pediatric epilepsy.
• Developing deep learning methods to explore neural correlates of gait in Parkinson’s Disease.
• Building a BCI pipeline for deep learning-based motor imagery applications using PyTorch, BCI2000, and MNE.

Data Science Intern

Dascena

• Evaluated deep learning models to predict C. Diff infection in the ICU, resulting in a journal publication.
• Built predictive models using longitudinal EMR data for multiple clinical indications.
• Developed EHR data processing pipelines using AWS and PySQL to enable analysis at scale.

Technical Operations Intern

Vertex Pharmaceuticals

• Developed a data acquisition and analysis pipeline for continuous drug production data.
• Built algorithms and a GUI for ETL and trend analysis on thousands of production reports.
• Increased analysis speed by 500x using Tabula, Pandas, PyQt, SQL, and SEEQ.

Vice President of Projects

Neurotech@Berkeley

• Developed a seamless EEG analysis and modeling platform for commercially available EEG devices.
• Created a data science-based curriculum to make neurotechnology accessible to the public.
• Built collaborations with neurotechnology research groups, leading to a student-led neurotech course at UC Berkeley.

Undergraduate Researcher

UC Berkeley, Maharbiz Group

• Designed novel automations and experiments to evaluate neural implant biocompatibility in silico.
• Assembled automated accelerated aging experiments using Raspberry Pi, PySerial, and MATLAB, resulting in a conference publication.

PhD

University of California, Los Angeles

Medical Informatics

B.S.

University of California, Berkeley

Bioengineering & Electrical Engineering & Computer Science

Graduate Research Fellowship Program (GRFP)

National Science Foundation (NSF)

Comparing P300 flashing paradigms in online typing with language models

PLOS ONE

Explores optimized P300 BCI flashing paradigms integrating language models for improved online typing performance.

Evidence of thalamocortical network activation during epileptic spasms: A thalamic stereotactic EEG study

Epilepsia

Demonstrates thalamocortical activation patterns during epileptic spasms using intracranial EEG.

Fast ripple band high-frequency activity associated with thalamic sleep spindles in pediatric epilepsy

Clinical Neurophysiology

Identifies a link between thalamic sleep spindle activity and high-frequency ripple dynamics in pediatric epilepsy.

Evaluation Of P300 Speller Performance Using Large Language Models Along With Cross-Subject Training

arXiv

Assesses the impact of large language models and cross-subject transfer learning on P300 BCI performance.

Uncovering spatiotemporal dynamics of the corticothalamic network at ictal onset

Epilepsia

Characterizes corticothalamic network dynamics during seizure onset using intracranial EEG.

Mentality: A Mamba-based Approach towards Foundation Models for EEG

ICLR 2024

Introduces a foundation model for EEG using Mamba-based state-space modeling.

High performance P300 spellers using GPT-2 word prediction with cross-subject training

Brain-Computer Interfaces

Improves P300 BCI typing speed using GPT-2 word prediction and cross-subject training strategies.

An Automated Deep Learning Approach for the Diagnosis of Limbal Stem Cell Deficiency

Investigative Ophthalmology & Visual Science

Develops an automated DL system for diagnosing limbal stem cell deficiency from ophthalmic imaging.

Pretrained transformers for seizure detection

ICASSP

Leverages pretrained transformers to improve seizure detection performance in EEG data. Achieved second place in seizure detection challenge.

MobileNetV2 based diagnosis and grading of limbal stem cell deficiency

IEEE BIBE

Presents a lightweight MobileNetV2-based DL model for limbal stem cell deficiency classification.

A comparative analysis of machine learning approaches to predict C. difficile infection in hospitalized patients

American Journal of Infection Control

Benchmarks ML approaches for predicting hospital-acquired C. difficile infection.

Retrospective validation of a machine learning clinical decision support tool for myocardial infarction risk stratification

Healthcare Technology Letters

Validates ML-based clinical decision support for MI risk prediction using EHR data.

Multicenter validation of a machine-learning algorithm for 48-h all-cause mortality prediction

Health Informatics Journal

Presents boosted-tree models for predicting 48h mortality using EHR data across multiple institutions.

Generation of model tissues with dendritic vascular networks via sacrificial laser-sintered carbohydrate templates

Nature Biomedical Engineering

Introduces a novel method for generating tissue models with perfusable vascular networks.

An automated system for reactive accelerated aging of implant materials with in-situ testing

IEEE EMBC

Develops an automated system for accelerated aging and in-situ evaluation of neural implant materials.