Senior Deep Learning Frameworks CUDA Software Engineer
Core
Integrate CUDA features and runtime abstractions into AI frameworks (PyTorch, JAX, vLLM, etc.) to enable high-performance, multi-GPU deep learning and HPC applications.
Role type
Senior IC CUDA software engineer (deep learning frameworks)
Builds
Production-grade AI toolkits, open-source framework integrations, and internal tools for distributed training and inference
Domain
High Performance Computing (HPC) and Artificial Intelligence (Deep Learning)
Deliverable
production ML models
Required skills
CUDA programming, C++, Python, deep learning frameworks (PyTorch, JAX), distributed machine learning, compiler technologies, performance profiling, HPC communication concepts, systems software fundamentals
Preferred skills
Expertise in deep learning autograd/training/inference internals, communication libraries (NCCL, MPI, UCX), kernel authoring (Triton, cuTe), deep learning compilers (Triton, XLA, torch.compile), compute/communication overlap programming
Technologies
CUDA, PyTorch, JAX, vLLM, TRT-LLM, SGLang, NCCL, MPI, UCX, Triton, XLA, torch.compile, Nsight Systems
Responsibilities
Integrate new CUDA features and runtime abstractions from PoC to production; Analyze AI workloads to identify innovation opportunities in lower stack layers; Drive improvements in the AI Compiler-Runtime interface for multi-GPU solutions; Design fault-tolerant and elastic solutions for large-scale AI workloads; Influence the core CUDA roadmap; Develop exploratory tools to profile and accelerate new deep learning paradigms; Write maintainable code for open-source and commercial releases
Seniority
Senior, hands-on IC
