Cross-Platform C++ AI Development with Conan, CMake, and CUDA

by Luis Caro Campos

From the talk at using std::cpp 2026:

Every year, the ISO C++ survey delivers the same verdict: dependency management is the #1 pain point for developers. For AI and machine learning development, introducing CUDA into the mix can cause that "pain point" to become a bigger bottleneck when developers these days are demanding “one-line, cross-platform” solutions.

This talk demonstrates how to use Conan and CMake to model the CUDA compatibility matrix directly in your code, achieving an ideal pipeline that works both on development machines, and CI, regardless of your target platform: One source checkout. One command. Identical builds on every platform.