A selection of research projects, tools, and explorations across computational neuroscience, brain-computer interfaces, and bioinformatics.
Investigating how world model representations and generated trajectories evolve across sequential reinforcement learning tasks. Training agents on a sequence of MiniGrid environments and semantically analyzing imagined trajectories from frozen model checkpoints to quantify retention and overwriting of prior task knowledge.
Developed a simulation framework to generate ground-truth RNA-seq datasets and benchmark transposable element quantification tools. Demonstrated that existing methods systematically overestimate TE expression; designed a modified reference transcriptome to correct for this inflation. Benchmarked across long-read (ONT, PacBio IsoSeq) and short-read platforms.
Contributed to data collection, signal processing, and classification for benchmarking custom low-cost zinc disc and concentric ring electrodes for wearable EEG. Published at IEEE EMBC 2024.
Using OpenBCI's Cyton board and EEG cap to play the NYT Spelling Bee game via SSVEP classification of visual cortex signals. A hybrid BCI integrating non-linear prediction models for high accuracy classification.
Led an EMG-controlled BCI project that navigates a computer using a custom armband and electrodes. Placed second in the International NeuroTechX Student Club Competition 2023.
Processed and prepared ImageNet stimulus data from the BOLD-5000 dataset to generate image predictions from fMRI neural data using OpenAI's DALL-E.
Contributed to three parallel sub-projects (motor imagery, P300 potential, and SSVEP) to build a hands-free communication system via EEG. Placed second in two international neurotechnology competitions.
Developed a QIIME2 plugin performing nested classification to identify the most phylogenetically similar animal gut microbiome to a given human sample. Built and trained scikit-learn classifiers at each node of a custom NCBI taxonomic tree.
Organized and taught course content to nearly a hundred students through weekly discussion sections and office hours. 100% student recommendation rate.
Advised and mentored student projects including Prosthetic Arm, Imagined Music EEG, Neurofeedback Software, Tetris with Motor Imagery, and P300 Potential Magician.
Curated materials to teach introductory Python and neural data analysis in a bi-weekly summer workshop.
Taught bioinformatics to high school students through SSHing to Raspberry Pis installed with bioinformatics pipelines.