Relational Foundation Model Engineer, Modern Data Stack

About NVIDIA and the Role

At NVIDIA, our focus is on advancing computing models dedicated to visual and AI computing. We are pioneers in visual computing and GPU technology, which has transformed the capabilities of computer graphics and enabled breakthroughs in deep learning applications such as robotics and autonomous vehicles. Currently, our Relational Foundation Model team is innovating a unified foundation model designed to interpret the structures and relationships within any relational database or heterogeneous graph, providing a novel approach to enterprise AI.

As an engineer on this cutting-edge team, you will be developing and experimenting with novel Transformer and Graph Neural Network (GNN) architectures that generalize across varied relational schemas. Your contributions will directly influence applications in recommendation systems, demand forecasting, fraud detection, and predictive maintenance through a single extensible model. Collaborating across the full machine learning lifecycle, you will engage in everything from architectural research and large-scale training to optimization and accelerating inference, pioneering foundational research with tangible production impact.

Key Responsibilities

• Partner with researchers and engineers to enhance Transformer and GNN models capable of handling any relational schema and heterogeneous graphs.
• Engage with vital real-world use cases including forecasting, entity matching, customer retention, and fraud detection using a unified foundation model.
• Apply your expertise in machine learning and AI to devise scalable and flexible solutions addressing practical challenges.
• Contribute across the entire machine learning system lifecycle, encompassing design, training, post-training optimization, and inference acceleration.
• Innovate and contribute to the evolution of NVIDIA’s next-generation Relational Foundation Model.

Required Qualifications

• Master’s or PhD degree in Machine Learning, Computer Science, or related field.
• Strong command of Python programming and experience with deep learning frameworks like PyTorch.
• A minimum of 8 years conducting research focused on ML algorithm development.
• Demonstrable practical experience applying predictive modeling techniques to real-world scenarios.

Preferred Attributes

• Experience or familiarity with graph-based machine learning techniques.
• Scholarly contributions published in leading conferences such as NeurIPS, ICLR, or ICML.

Additional Information

NVIDIA is recognized as one of the most sought-after employers internationally, boasting a culture of innovation and autonomy. We value creativity and independent thinking, striving for excellence in AI and computing technologies.