Senior Compute Kernel Architect, GPU Power

The Role: Stress-Testing the Future of AI

In this position, you won't just be optimizing for speed; you'll be writing code designed to push hardware to its absolute breaking point. As a Senior Compute Kernel Architect, you will develop specialized CUDA kernels that act as 'stress workloads,' intentionally triggering extreme power draw and voltage fluctuations. Your goal is to validate the Power Delivery Network (PDN) of next-generation GPUs, ensuring that NVIDIA's silicon remains stable under the most demanding real-world AI and compute scenarios.

Technology & Vibe Coding Integration

This role sits at the fascinating intersection of low-level assembly and modern AI-assisted development. While you'll be deep in the weeds with C++, PTX, and di/dt analysis, NVIDIA encourages the use of 'vibe coding'—leveraging LLMs and AI tools—to handle the Python-based automation, characterization scripts, and complex data analysis pipelines.

• Key Tech: CUDA, C++, PTX, Nsight Systems, and Python.
• AI Workflow: Use natural language prompting to rapidly iterate on the scripts that monitor and analyze hardware telemetry.

Why This Matters

You will be working at the frontier of accelerated computing, directly influencing the physical architecture of the chips that power the global AI revolution. This is an opportunity to bridge the gap between software logic and silicon reality, using the latest AI development methodologies to build more resilient hardware.