Research Intern, Models and Synthetic Data Autonomous Driving - Summer

Be part of the Autonomous Vehicle (AV) Applied Research team and help advance safe autonomy through generative world models and synthetic data. This internship is an opportunity to work alongside a team of researchers and engineers developing the next generation of models that can understand, simulate, and reason about complex driving environments. We are looking for a passionate and highly motivated Research Intern with a strong background in machine learning and an interest in autonomous driving, generative world models, multimodal foundation models, and synthetic data generation.

Our applied research team works closely with researchers and engineers across NVIDIA Research and the AV organization to develop scalable methods for training, evaluating, and improving autonomous systems. Our research spans generative video and world models, controllable scenario generation, multimodal learning, end-to-end driving, 3D computer vision, and data-centric learning. We are particularly interested in using generative models to create diverse, safety-critical, and long-tail driving experiences that can help improve the robustness and generalization of autonomous systems.

As a Research Intern, you will contribute to research on generative data engines for autonomous driving: developing and using world models to understand data coverage, generate targeted driving scenarios, and study how synthetic data can systematically improve AV models. You will have the opportunity to develop new research ideas, run large-scale experiments, collaborate with experienced researchers and engineers, and contribute to publications at top-tier conferences.

• Research and prototype generative world models, traffic world models, and synthetic-data generation methods for autonomous driving.
• Develop traffic world models that capture multi-agent interactions and realistic, reactive behaviors, enabling simulation and generation of complex driving scenarios.
• Develop methods for controllable generation of diverse, interactive, and safety-critical driving scenarios.
• Explore how generative video models, multimodal foundation models, traffic simulation, data curation, active learning, self-supervised learning, and simulation can improve training-data quality, diversity, and coverage.
• Design and conduct experiments to understand how generated data affects AV model performance, robustness, and generalization.
• Translate recent advances in generative modeling and machine learning into practical approaches for autonomous driving.
• Collaborate closely with researchers and engineers across NVIDIA on research prototypes and experimental systems.
• Contribute to high-impact research publications and open research efforts.

• Currently pursuing a Ph.D. or M.S. in Computer Science, Electrical/Computer Engineering, Robotics, or a related field, with research experience in deep learning, computer vision, generative modeling, multimodal learning, robotics, or autonomous driving.
• Strong foundations in machine learning and deep learning, with hands-on experience developing and evaluating neural-network models.
• Experience in one or more of the following areas: generative models, video generation, multimodal learning, 3D computer vision, autonomous driving, or robotics.
• Strong mathematical and analytical skills and an interest in designing rigorous experiments.
• Strong Python programming skills and experience with modern deep learning frameworks.
• Self-motivated, collaborative, and comfortable working in a research environment with open-ended problems.

• Publications or research projects at conferences such as NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, or related venues.
• Hands-on experience with…