Software Engineer - Model Performance Systems
Toronto, ON +1HybrideTemps plein
Baseten
Candidates need Python skills, interest in GPU and networking systems, and an automation mindset. Strong communication, the ability to work with vague requirements and mentoring experience are also expected. C++ is a plus.
PyTorch Profiler, NVIDIA Nsight Systems, py-spy, GitHub Codespaces, InferenceMAX, genai-bench
Software Engineer - GPU Kernels
Toronto, ON +1HybrideTemps plein
Baseten
Candidates need a strong understanding of GPU architecture and memory hierarchy, proficiency in C++ and GPU profiling tools, and knowledge of the CUDA C++ API, numerical precision and modern GPU features. Experience with Transformer models, Flash Attention, Cutlass, Triton or GEMM tuning is a plus.
CUDA, PTX, Nsight Systems, Nsight Compute, Torch Profiler, Cutlass