There is an increasing need to bring machine learning to a wide range of hardware devices. Current frameworks rely on vendor-specific libraries and optimize for a narrow range of server GPUs. Deploying workloads to new platforms, such as mobile phones, embedded devices, and accelerators, requires significant manual effort. Because of this, companies are opting to choose vendors' downstream kernels and userspace. This means that they are less likely to benefit from Collabora's work on upstream components. In this talk, we will walk through the process of building a computer vision pipeline on top of a completely open-source stack: open-source GPU driver, machine learning framework, and models.