Saarthak Sarup
Computational Neuroscience · Neuromorphic Computing
I study the architecture of neural circuits. I also paint.
Both are attempts to understand structure.
Selected Work
Synaptic Clustering in Mouse Visual Cortex
Discovered universal synaptic clustering patterns across petabyte-scale connectome datasets, revealing ensemble-specific wiring that challenges classical models of dendritic integration.
Read more →
Connectomics MCP Server
An open-source library wrapping neuPrint, MICrONS, and FlyWire behind the Model Context Protocol, enabling LLM agents to query connectomic databases in natural language.
View on GitHub →
Brian2 Models MCP Server
An open-source library exposing a searchable catalog of Brian2 neural simulation models via the Model Context Protocol, enabling LLM agents to find, retrieve, and adapt published simulation code in natural language.
View on GitHub →
Analyzing Axodendritic Motifs in Petabyte-Scale Connectomes
Developed efficient hypergraph mining algorithms that reduced complexity from quadratic to log-linear, enabling motif discovery at scale across the largest available connectome reconstructions.
Read more →
Field Guide for Parks
An offline and local AI-powered nature guide for national parks. Runs a multimodal model entirely on-device to identify plants, animals, and geological features without cell service.
Learn more →Art
A small collection of paintings — mostly acrylics and oil — made between lab hours and late nights.
