Python/C++ Engineer | Flask, Microservice, API, NumPy, pybind11, CMake

We're building 3-D medical images, voxel grids in real millimetre space, moved by rigid transforms, sampled between voxels. Python on top, C++ underneath where it has to be fast. - Python APIs callers can't misuse, coordinate spaces that don't get mixed up, failures that can't be ignored - C++ per-voxel work: interpolation, resampling, gradients. Correct at the boundaries, fast in the middle. - Hunt the one-dataset bug: off-by-one at an edge, float that should've been a double, index type that overflows on a big volume - Write the tests. No QA team behind you.