CUDA Kernel Engineer for AI Hardware & Semiconductors

Voltai; Palo Alto (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Voltai is building world models and AI-enabled semiconductor tools. You will develop, integrate, and optimize CUDA kernels to accelerate design and verification tasks across thousands of GPUs.

You will create tooling, benchmarks, and integration layers, collaborating with researchers to advance AI-driven hardware design and release kernels to open-source ecosystems. Youll work closely with researchers and engineers to push the limits of GPU utilization for compute-intensive workloads and

Qualifications

• Writing and optimizing CUDA kernels for large-scale AI workloads.
• Profiling and optimizing GPU performance for compute-heavy tasks.
• Integrating custom kernels into PyTorch, Megatron, vLLM, TorchTitan.
• Working with NVIDIA hardware and software stacks (Hopper/Blackwell, NVLink, NCCL, Triton).
• Building GPU-accelerated primitives for graph reasoning and hardware sim.
• Collaborating with AI researchers and semiconductor experts to translate workloads into high-performance GPU code.

Responsibilities

• Develop, integrate, and optimize state-of-the-art CUDA kernels for AI models.
• Power large-scale model training, inference, and reinforcement learning workloads.
• Build tools, benchmarks, and integration layers to maximize GPU utilization.
• Collaborate with researchers and engineers to push Voltai's AI+semiconductor objectives.

Skills

CUDA kernels

GPU profiling

Framework integration

NVIDIA hardware

GPU primitives

Research collaboration

Tools

PyTorch

Megatron

vLLM

TorchTitan

NCCL/Triton

Job description

Voltai is building world models and AI-enabled semiconductor tools. You will develop, integrate, and optimize CUDA kernels to accelerate design and verification tasks across thousands of GPUs.

You will create tooling, benchmarks, and integration layers, collaborating with researchers to advance AI-driven hardware design and release kernels to open-source ecosystems. Youll work closely with researchers and engineers to push the limits of GPU utilization for compute-intensive workloads and