Loading signal
Preparing the next layer of the atlas.
builds
BCI & Real-Time Systems
Real-time neural data streaming, processing, classification, and foundation-model experimentation framework.
Modular orchestration layer built on top of NeuroForge for BCI pipeline development.
Mechanistic Interpretability
Mechanistic interpretability tools for neural multi-modal transformers.
System of Agents to generate BCI middleware code.
Neural Data Infrastructure
DataJoint Element for behavioral analysis with DeepLabCut - contributed improvements.
DataJoint Element for multi-photon calcium imaging analysis.
DataJoint Element for facial inference and video-derived behavior features with Facemap.
Led and implemented cloud-ready DataJoint workflows for multimodal neuroscience data, including electrophysiology, calcium imaging, fiber photometry, DeepLabCut pose estimation, Facemap facial inference, behavior, and visualization across large customer deployments.
Led a major customer deployment for Harvard/Sabatini Lab involving multimodal neuroscience processing and visualization over ~12TB-scale datasets.
Contributed to large-scale neuroscience data workflow design and deployment for Allen Institute Mindscope, supporting large multimodal datasets and cloud-oriented processing patterns.
Implemented DeepLabCut pose estimation and Facemap facial inference model training/inference into DataJoint’s open-source framework and deployed both in cloud architectures for Lu Lab at Indiana University.
Scientific Workflow Systems
Contributed to DataJoint’s open-source Element ecosystem and helped create standard data processing templates for electrophysiology, calcium imaging, DeepLabCut, Facemap, and related neuroscience modalities.
Electrophysiology Infrastructure
Worked on integrating SpikeInterface-style electrophysiology processing into DataJoint’s array electrophysiology ecosystem, helping connect spike sorting workflows to structured database schemas.
Scientific DevOps
Built or contributed to a workflow monitoring system to manage different customer projects and track errors across distributed DataJoint workflows.
NEATLABs Research
Served as Lead Lab Programmer at NEATLABs, building behavioral paradigms, large-scale LFP/behavior analysis pipelines, and neural engineering analyses for decision-making, inhibition, TBI, and reward-related neural networks.
Applied Stanford dynamic time warping algorithms and Tensor Component Analysis to multi-terabyte LFP datasets, identifying precise temporal neural patterns.
Real-Time Creative Neurotech
Built real-time LED strip visualization systems connected to NeuroSky-style brain input and PC audio output, multiplexed through Raspberry Pi, speakers, and headphones for synchronized visual feedback.
Neural Decoding and ML
Created a neural spike data augmentation method to expand limited training samples and improve Kalman filter-based decoding of arm position from M1 neural activity.
AWS-based 2D FFT processing pipeline for precision neuroscience.
Personal Product Experiments
Pet fitness tracking and social platform inspired by Shasta.
ECoG and EEG deep learning based signal classification package.
Applied AI Products
Helped build backend and frontend for CoEval, a nonprofit web platform deployed on Cloudflare Workers to organize a database of scientific AI companies and allow selected users to vote on leading AI practices.
Audio-Visual Systems
Worked on Dolby Vision HDR picture quality testing, white-point balance testing, and glasses-free 3D television quality analysis, including training senior engineers on updated testing processes.
DataJoint Element for optogenetic experiments
DataJoint Element for miniscope calcium imaging analysis with CaImAn
DataJoint Element for localizing Neuropixels electrodes
DataJoint Element for Animal Management - NIH U24
DataJoint Element for Lab Management - NIH U24
Example DataJoint workflow for behavioral analysis with DeepLabCut
DataJoint Element for event-based behavior experiments
DataJoint Element for Extracellular Array Electrophysiology - NIH U24
Common functions for the DataJoint Elements
Bioengineering Hardware
Designed biomedical signal and implant-adjacent systems including an intracranial pressure monitor concept and EMG Morse-code decoder with continuous data acquisition and spike/signal analysis in MATLAB.
DataJoint Element for Session Management - NIH U24
Example DataJoint workflow of `element-array-ephys` - NIH U24
A general purpose repository containing all generic tools/utilities surrounding the DataJoint ecosystem. These can be candidates for DataJoint plugins
DataJoint Elements is a collection of curated computational workflows for neurophysiology experiments.
A docker image optimized for running a JupyterLab environment with DataJoint Python.
DataJoint Archive Client
A minimal base docker image with DataJoint Python dependencies installed.
Biomedical ML
Completed multiple biomedical ML and computational biology projects spanning single-cell RNA-seq, bulk RNA-seq, ChIP-seq, tumor MRI classification, pharmacokinetic modeling, and depth-map CNN classification.
Cloud Infrastructure
Precision Neuroscience Interview 2DFFT AWS pipeline
Neuroscience Research
Applied Kalman Filter on single motor neuron activity in M1 cortex to predict arm trajectory. Applied data augmentation to original dataset to decreased error by 18.5%.
MNE: Magnetoencephalography (MEG) and Electroencephalography (EEG) in Python
Modules for processing extracellular electrophysiology data from Neuropixels probes
A Python-based module for creating flexible and robust spike sorting pipelines.
Neo is a package for representing electrophysiology data in Python, together with support for reading a wide range of neurophysiology file formats
A minimal base docker image with conda.