GSOC 2026 Project #31 : Accelerating Virtual Brain Inference from Neuroimaging

GSoC
1

Skill level: Intermediate to Advanced

Required skills: Data analysis, Simulations of differential equations, Python and Git; familiarity with JAX, whole-brain models, and simulation-based inference would be beneficial.

Time commitment: Full time (350 hours)

About: Virtual Brain Inference (VBI) provides fast simulations, taxonomy of feature extraction, efficient data storage and loading, and probabilistic machine learning algorithms, enabling biophysically interpretable inference from non-invasive and invasive recordings. Scalable JAX simulations and automatic feature extraction will support the use cases.

Aims:

  • Reproduction of existing use cases in JAX

  • Incorporating scalable JAX-based simulations

  • Automatic feature extraction in the use cases

  • Tuning and testing model parameters and in silico-validation

Final output will be a lightweight demonstrator with clear documentation, enabling users to quickly run a standardized example. Website: GitHub - ins-amu/vbi: Virtual Brain Inference ยท GitHub

2

Hi @mhashemi,

I am Hengye Zhu, an undergraduate student interested in High-Performance Computing for Computational Neuroscience. Although my previous experience has centered on micro-scale neuronal models, I have always been deeply interested in whole-brain modeling.

I have a background in Python/C++ and have taken coursework in stochastic processes, ODEs, and PDEs. Would starting with this issue be an appropriate way for me to become familiar with VBI?

My background is as follows:

I am currently an editor for NeuroML. I contributed a PR to GeNN and participated in GSoC 2025. Additionally, I led the front-end modeling for a biophysically detailed neuron simulator (currently under development) and participated in building a JAX-based automatic differentiation engine for the simulator.

Best regards,

Hengye Zhu