Senior AI Research Engineer - GPU Systems & Kernels
Core
Partner with Research Scientists to optimize large-scale model training and inference speed and efficiency at the hardware level.
Role type
Senior IC GPU systems engineer (LLM training/inference)
Builds
High-performance infrastructure for training and serving large language models, including simulation environments for Trained Agents.
Domain
Cloud observability, security, and large-scale AI model training
Deliverable
production ML models
Required skills
GPU kernel development and optimization (CUDA, Triton), low-latency communication and collective operations (RDMA, NCCL), systems programming (C++, Rust, CUDA C), Python, hardware constraint reasoning (memory hierarchy, tensor cores, interconnect topology), profiling and debugging (Nsight)
Preferred skills
Mixture-of-Experts models, expert parallelism, sparse model inference, GPU communication libraries (NVSHMEM), kernel frameworks (CUTLASS), distributed training frameworks (Megatron-LM, TorchTitan, DeepSpeed), GPU-intensive systems in graphics/rendering/simulation
Responsibilities
Write, tune, and debug GPU kernels for LLM operations; Optimize for latest NVIDIA GPU generations; Reduce communication costs in distributed workloads; Profile end-to-end training workloads; Build benchmarks, profiling tools, and regression tests; Collaborate with Research Scientists to implement new model architectures; Explore emerging techniques like low-precision training and speculative decoding
