Senior LLM Agents Architect
Core
Design and build autonomous LLM-driven systems that reason about hardware, write high-performance CUDA, and automate architectural simulation, analysis, and optimization for NVIDIA GPUs.
Role type
Senior IC LLM Agents Architect (GPU/Hardware)
Builds
End-to-end agent flows for kernel optimization, automated performance forensics, and GPU architectural studies.
Domain
Semiconductor hardware design, GPU architecture, and applied AI/ML systems.
Deliverable
production ML models
Required skills
CUDA programming, computer architecture (memory hierarchies, parallelism), Python, C++, LLM agent orchestration, RAG pipelines, performance profiling (Nsight), system design, observability.
Preferred skills
PyTorch compilation stack (TorchInductor, Triton), HPC performance engineering, distributed processing (NVLink, InfiniBand), frontier agentic coding tools.
Responsibilities
Design agentic AI systems to generate and optimize GPU compute kernels; collaborate with GPU architects to encode domain expertise into agent workflows; build automated performance forensics agents; develop agentic flows for GPU architectural studies; prototype and productize solutions; set up evaluation backbones; mentor teams on agent orchestration and observability.
Seniority
Senior, hands-on IC
