5 days ago
Responsibilities
- Develop, train, and evaluate models for accurate, temporally consistent reconstruction of dynamic scenes.
- Create architectures for Gaussian prediction, neural rendering, 3D and 4D reconstruction, object-centric representations, and mapping.
- Model geometry, appearance, semantics, motion, and interactions for realistic and controllable simulation environments.
- Explore diffusion, flow-based, and video-generation methods for novel views, scene completion, temporal prediction, and world generation.
- Scale data and distributed training pipelines for multi-camera video, vehicle poses, perception signals, and other sensor data.
- Build visualization and analysis tools to understand model behavior and guide measurable improvements.
- Integrate trained models into simulation workflows with production teams to improve reliability and efficiency.
Requirements
- At least five years of relevant experience and a BS, MS, or PhD in a related technical field, or equivalent practical experience.
- Proficiency in Python and experience developing and training models with PyTorch or a comparable framework.
- Strong foundation in deep learning, computer vision, 3D geometry, multi-view geometry, neural rendering, or generative modeling.
- Experience training and evaluating models with large image, video, 3D, or multimodal datasets.
- Ability to work with camera models, calibration, coordinate systems, geometry, motion, uncertainty, and temporal consistency.
- Experience improving models for noisy data, dynamic objects, occlusions, incomplete observations, and uncommon scenarios.
- Systematic approach to debugging, metrics, controlled experiments, and model failure analysis.
- Clear communication and ability to collaborate across research and production engineering teams.
- Preferred experience with Gaussian splatting, feed-forward 3D reconstruction, NeRFs, differentiable rendering, neural scene representations, dynamic reconstruction, diffusion models, flow matching, video generation, world models, novel-view synthesis, generative simulation, autonomous driving, robotics, simulation, synthetic-data generation, distributed training, CUDA optimization, or GPU profiling.
- Publications, open-source contributions, or production results in related fields are beneficial.
Categories
AI ResearchML Engineering
About Nvidia
Since its founding in 1993, NVIDIA (NASDAQ: NVDA) has been a pioneer in accelerated computing. The company’s invention of the GPU in 1999 sparked the growth of the PC gaming market, redefined computer graphics, ignited the era of modern AI and is fueling the creation of the metaverse. NVIDIA is now a full-stack computing company with data-center-scale offerings that are reshaping industry.