Research Scientist, AI for Graphics and Gaming -… at NVIDIA · Santa…

Summary: NVIDIA is seeking a Research Scientist in Generative AI for Graphics and Gaming to join their Applied Deep Learning Research team. The role involves conducting cutting-edge research, training advanced models, and prototyping in real-time engines to improve graphics quality and deliver transformative technology in gaming applications. Responsibilities: - Research and develop AI models that improve real-time graphics quality, robustness, performance, and latency, and enable new interactive experiences such as world models, LLM-driven gameplay, and AI-driven characters - Design and run large-scale pre-training and post-training for foundation models for graphics and gaming, using NVIDIA’s GPU infrastructure for experimentation and iteration - Build and own data pipelines and datasets for foundation models and large-scale training, spanning synthetic data, in-engine captures, and real-world content - Develop and prototype models for real-time image, video, and 3D content generation in modern game engines and graphics frameworks, and create demos that clearly showcase player-visible benefits - Stay current on AI and graphics research and work with product, driver, hardware, engine, and platform teams to turn promising ideas into shipping, user-visible features Required Qualifications: - Pursuing PhD in Computer Science/Engineering, Electrical Engineering, Applied Mathematics, or a related field (or equivalent experience) - Deep learning experience in image/video generation, graphics, computer vision, or related fields, with a solid understanding of core DL fundamentals - Hands-on experience training large models on multi-GPU clusters (hundreds to thousands of GPUs) - Building robust distributed training pipelines and debugging at scale - Strong proficiency in PyTorch required - Proven track record of impactful research and shipped ideas (e.g., leading publications, production features, or widely used tools) - Strong communication and collaboration skills in cross-functional teams Preferred Qualifications: - Familiarity with GPU architecture, CUDA abstractions, and efficiency techniques (e.g., distillation, pruning, low-precision training/inference) - Background in graphics—such as rendering pipelines or game engines - Experience with generative models, neural rendering, or reconstruction Required Skills: Deep Learning, Image Generation, Video Generation, Computer Vision, PyTorch, Distributed Training, GPU Architecture, CUDA, Rendering Pipelines, Game Engines, Generative Models, Neural Rendering Benefits: Equity, Benefits