Software Engineer, ML Performance
Core
Optimize performance of distributed AI training and inference workloads across the entire stack (model, framework, compiler, CUDA, network) for NVIDIA's GPU solutions.
Role type
Senior IC software engineer (ML performance)
Builds
High-performance GPU training and inference systems for AI workloads
Domain
AI / Accelerated Computing / GPU Systems
Deliverable
production ML models
Required skills
GPU workload optimization, distributed systems analysis, performance bottleneck identification, regression analysis, automated diagnosis evaluation, Python, CUDA
Preferred skills
PyTorch, JAX, Megatron, NCCL, TensorRT-LLM, vLLM, SGLang, Dynamo, counter-factual analysis, debugging tool development
Technologies
CUDA, NCCL, PyTorch, JAX, Megatron, TensorRT-LLM, vLLM, SGLang, Dynamo
Responsibilities
Analyze performance of distributed training and inference workloads across the stack; Define performance metrics tracked across a fleet of GPUs; Drive analysis of regressions to isolate causes; Evaluate automated diagnoses for accuracy; Work with build partners to carry findings to improvements
Seniority
Senior, hands-on IC
