Solutions Architect - Deep Neural Network Evaluation

NVIDIA’s Worldwide Field Operations (WWFO) team is looking for an AI focused Solution Architect with expertise in evaluation of neural networks and agentic pipelines. A candidate with understanding of modern evaluation datasets and tools allowing to work with key model builders / AI application developers across EMEA to improve core models as well as end to end agentic pipelines. As Solutions Architect, you will be the first line of technical expertise between NVIDIA and our customers. Your responsibilities will vary from working on proof-of-concept demonstrations, to driving relationships with key executives and managers to promote NVIDIA technologies. This role focuses on the evaluation, analysis, and design of agentic AI systems, with a strong emphasis on LLMs and retrieval-based architectures. It involves benchmarking AI systems across diverse use cases and languages, evaluating performance across the end-to-end stack from LLMs and VLMs to embedding models. The position also requires deep analysis of system failures and the development of mitigation strategies, including model selection, retrieval pipeline optimization, and LLM or embedding model fine-tuning. We are looking for a candidate passionate about AI with strong expertise in neural network development and fine-tuning. The ideal candidate will be able to collaborate across customers, engineering teams, industry business development, and corporate marketing. In this role, you will work with the latest LLM, VLM, and embedding technologies that are redefining how people interact with intelligent systems. What You’ll Be Doing: Work directly with key customers to understand their technology and provide the best solutions. Drive adoption of NVIDIA open-source evaluation tools (NeMo) and help the community to develop standard methodology for model's robustness measurement. Develop and demonstrate solutions based on NVIDIA’s and open-source LLM technology and integrate them into agentic pipelines. Partner with e...