Senior Python developers. (Python, PyTorch, CUDA)

They offer virtualized GPUs accessible over a network, enabling users to utilize cloud GPUs as easily as network-attached storage. It allows developers to access and scale GPU resources on demand, only paying for active use. With an emphasis on simplicity, cost efficiency, and instant scaling, they target developers and enterprises needing GPU power for AI, data science, or prototyping, eliminating the complexity of manual GPU management. The solution integrates seamlessly into existing workflows, helping teams boost productivity and reduce hardware costs.

1-2 rounds of interview (test assignment can be among those).

Looking for a team of engineers (2 seniors to start with) for the next 2-3 months to build a repo of test scripts in Python. These scripts would cover a wide range of content, largely data science and AI workflows requiring GPUs. They want a mix of PyTorch, CUDA, and other test scripts that can all easily be run from the same place. The goal is to test their GPU virtualization solution as thoroughly as possible from every angle.

The goal is to help pressure test their system, especially models that use multiple GPUs. The task at hand would require building less like conventional tests but more like replicating data science workloads (e.g., running different models, like CUDA kernels, testing things with multi threading, profiling, etc.).

Need to know Python, PyTorch, CUDA.