Track emerging research trends and turn gaps between researcher needs and NVIDIA's o erings into prototypical solutions and direct feedback to NVIDIA Engineering.
Maintain deep expertise in your domain while staying versatile across NVIDIA's full platform: GPUs, CPUs, networking, and software
What we need to see
A graduate degree from a leading university in a STEM related discipline.
5+ years of hands-on experience running the large language model lifecycle across multi-node systems: training, fine-tuning, inference, serving, and/or agentic workflows.
Strong collaboration and communication skills, with the ability to build relationships with academic and research stakeholders, and to communicate complex ideas clearly to both expert and non-expert audiences.
Action oriented, analytical, self-motivated, and a passionate learner, with excellent organization skills to work in a heavily multi-tasked environment.
Fluent in English, both oral and written, and comfortable working in Python.
Ways to stand out from the crowd
A PhD from a leading university in a STEM related discipline, with 3+ years of research on large language models or foundation models and their applications.
A track record of thought leadership: high-impact publications, talks at academic conferences and workshops, as well as good understanding of scientific policy engagement, grant processes, or national/European research program structures.
Experience with NVIDIA's AI software stack, powered by CUDA and CUDA-X libraries, e.g. NVIDIA AI Enterprise, NeMo Framework, Megatron Bridge, NIM, TensorRT-LLM, Dynamo, NeMo Agent Toolkit, and Triton Inference Server, as well as the Nemotron open-model methodology.
As a Solutions Architect, there is travel involved (30% of the working time), as often the best way to figure things out is a face to-face meeting, but the job is not life on the road. We make heavy use of conferencing tools, and you are empowered to figure out how to get the job done and do what it takes to make our customers successful.