komamripy - Python Wrapper for KomaMRI

KomaMRI.jl is a high-performance Julia package for MRI simulation and differentiable workflows, but its use is currently limited to the Julia ecosystem, creating a barrier for the broader Python-based scientific computing and machine learning community. This project introduces komamripy, a lightweight Python wrapper that exposes KomaMRI’s core simulation and differentiable capabilities through a minimal, intuitive API. The approach leverages Julia–Python interoperability to map data structures between environments, incorporates Julia-side precompilation to reduce first-use latency, and enables flexible GPU backend support across CUDA, AMDGPU, Metal, and oneAPI. The project will deliver a pip-installable Python package with CI-tested builds and automated releases, precompiled simulation and autodiff workflows, cross-platform GPU backend selection, and a validated end-to-end example demonstrating differentiable RF pulse optimization using JAX, along with supporting documentation and example scripts.