Portfolio Roadmap

Matthew
Tarchick

Ph.D. · Building biological signal systems.

I work at the intersection of experimental neuroscience, signal processing, mechanistic modeling, and scientific software.

I am building toward tools that connect living biological recordings to models that explain them. My work starts with real tissue and noisy signals, then turns those signals into analysis pipelines, simulations, biomarkers, and reusable software.

github.com/mattar13 Email Google Scholar LinkedIn Resume PDF
The direction
01
Measure the tissue
Patch-clamp, ERG, calcium imaging, dopamine sensors, two-photon/confocal microscopy.
02
Extract the signal
Pipelines for neural time-series, retinal images, electrophysiology, oculomotor data, and noisy biological recordings.
03
Build the model/tool
Mechanistic models, digital twins, biomarkers, scientific packages, and hardware/software systems.
Showcase
Built pieces of the roadmap
Signal extraction
ElectroPhysiology.jl
↗
A toolkit for loading, analyzing, and visualizing electrophysiology recordings. It turns high-volume retinal and neural time-series into reproducible spike, burst, wave, and ERG analyses. It proves I can turn noisy biological recordings into reusable scientific software.
JuliaElectrophysiologySignal ProcessingReusable Tools
VascularAnalysis
↗
A pipeline for quantifying retinal vasculature from 3D confocal image stacks. It converts microscopy volumes into vessel maps and measurable vascular features. It proves I can connect biological imaging to scalable computer-vision analysis.
PythonCellPoseComputer Vision3D Imaging
Oculomotor biomarker pipelines
Eye-tracking workflows for extracting movement features from phone and computer video. They make low-cost video usable for physiological and neurological signal analysis. They prove I can build translational pipelines from consumer sensors to candidate biomarkers.
PythonComputer VisionBiomarkers
Biological modeling
RetinalChaos
↗
A computational model of cholinergic retinal wave generation, propagation, and direction bias. It connects developmental retinal circuit behavior to interpretable dynamical mechanisms. It proves I can build models that stay accountable to experimental recordings.
JuliaNetwork ModelsRetinal WavesPublished
RetinalTwin.jl
↗
A biophysically grounded simulation of retinal circuit activity and ERG-like outputs. It links cell-level parameters to tissue-level electrical behavior. It proves I can design model architectures that connect physiology, parameters, and observable signals.
JuliaODEBiophysicsDigital Twin
VesselDigitalTwin
↗
A computational model of retinal vascular network formation and remodeling. It explores how local vessel dynamics can produce tissue-scale vascular structure. It proves I can combine agent-based and diffusion models around biological mechanisms.
PythonPDEAgent-BasedDigital Twin
Translation / tools
muERG
↗
A 3D-printed ex vivo electroretinography platform for isolated retinal tissue. It makes tissue electrophysiology more accessible, repeatable, and scalable. It proves I can build across hardware, physiology, analysis, and translational constraints.
CADElectrophysiology3D PrintingmuERG
protein-seq-to-func
A retrieval pipeline that links protein sequence information to functional literature. It helps turn biological text and sequence context into a searchable functional knowledge base. It proves I can connect scientific literature, embeddings, retrieval, and domain questions into usable tools.
PythonRAGProteinsLive App
GitHub ↗ Live app ↗
dopamine imaging pipeline
↗
An analysis workflow for infrared dopamine sensor imaging on adapted microscopy systems. It extracts dynamic neuromodulatory signals from challenging optical recordings. It proves I can move from custom biological measurement to interpretable signal analysis.
DopamineImagingSignal Extraction
Current Focus
What I am building now
Eye-tracking and oculomotor biomarkers
Eye-tracking and oculomotor biomarkers from phone/computer video.
VideoBiomarkersOculomotor
Physiological state estimation
Physiological state estimation from EEG, HRV, and wearable signals.
EEGHRVWearables
Retina-centered disease models
Retina-centered models that connect cells, circuits, vasculature, and disease.
RetinaDiseaseDigital Twin
Thinking
How I think about the work
I like models that are accountable to biology. A model should know where its parameters came from. A signal-processing pipeline should preserve the shape of the original recording. A biomarker should trace back to a physiological mechanism. A tool should make the next experiment easier.
Evidence
Compact proof points
Training and research
Ph.D. in Integrated Bioscience · Postdoctoral research at UC Berkeley · Publications in retinal development, disease, electrophysiology, and computational modeling
NeuroscienceRetinaModeling
Building record
Co-inventor of the muERG electrophysiology platform · Built open-source tools in Python and Julia · Experience with biological recordings, imaging, modeling, and translational signal analysis
HardwareScientific SoftwareSignal Analysis