Nvidia Senior DevTech Engineer - Windows LLM GenAI Open-Source Ecosystem 

Germany, North Rhine-Westphalia 

586257752

For our team in Wuerselen we are now looking for a Developer Technology Engineer to ...

• contribute to the LLM & GenAI open-source ecosystem to enable Windows AI enthusiasts and developers with innovative models and functionality as well as speed-of-light performance on RTX.

• engage with our strategic partners and internal teams to overcome the challenges arising when deploying modern LLM & GenAI architectures on local workstations.

What you’ll be doing:

• Improve Windows LLM & GenAI user experience on NVIDIA RTX by working on feature and performance enhancements of OSS software, including but not limited to projects like PyTorch, llama.cpp, ComfyUI.

• Engage with internal product teams and external OSS maintainers to align on and prioritize OSS enhancements.

• Work closely with internal engineering teams and external app developers on solving local end-to-end LLM & Generative AI GPU deployment challenges, using techniques like quantization or distillation.

• Apply powerful profiling and debugging tools for analyzing most demanding GPU-accelerated end-to-end AI applications to detect insufficient GPU utilization resulting in suboptimal runtime performance.

• Conduct hands-on trainings, develop sample code and host presentations to give good guidance on efficient end-to-end AI deployment targeting optimal runtime performance.

• Guide developers of AI applications applying methodologies for efficient adoption of DL frameworks targeting maximal utilization of GPU Tensor Cores for the best possible inference performance.

• Collaborate with GPU driver and architecture teams as well as NVIDIA research to influence next generation GPU features by providing real-world workflows and giving feedback on partner and customer needs.

What we need to see:

• 5+ years of professional experience in local GPU deployment, profiling and optimization.

• BS or MS degree in Computer Science, Engineering, or related degree.

• Strong proficiency in C/C++, Python, software design, programming techniques.

• Familiarity with and development experience on the Windows operating system.

• Proven theoretical understanding of Transformer architectures - specifically LLMs and Generative AI - and convolutional neural networks.

• Experience working with open-source LLM and GenAI software, e.g. PyTorch or llama.cpp.

• Experience with CUDA and NVIDIA's Nsight GPU profiling and debugging suite.

• Strong verbal and written communication skills in English and organization skills, with a logical approach to problem solving, time management, and task prioritization skills.

• Excellent interpersonal skills.

• Some travel is required for conferences and for on-site visits with external partners.

Ways to stand out from the crowd:

• Experience with GPU-accelerated AI inference driven by NVIDIA APIs, specifically cuDNN, CUTLASS, TensorRT.

• Confirmed expert knowledge in Vulkan and / or DX12.

• Familiarity with WSL2, Docker.

• Detailed knowledge of the latest generation GPU architectures.

• Experience with AI deployment on NPUs and ARM architectures.