Intel · posted · checked
AI RnD Engineer, XeSS and Neural Graphics (f/m/d)
PythonPytorchHybrid · MunichSenior
Who can apply
Hybrid · Munich
Location: Germany, Munich
Job Details:
Job Description:
Drive the development of AI models for Intel XeSS and related AI-based graphics technologies for Client Graphics, with direct impact across XeSS Super Resolution, XeSS Frame Generation, neural rendering, and next-generation AI rendering capabilities. You will contribute to technical direction and drive execution across research, SW, HW architecture, validation, and ecosystem teams to bring new technologies from concept to product.
This is an end-to-end technical role spanning model design, datasets, training, visual quality, performance optimization, and product integration. You will help shape how modern AI model architectures are adapted for the quality, performance, and latency constraints of real-time graphics.
Key Responsibilities:
Drive the development of AI models for XeSS and next-generation neural rendering capabilities, from early exploration through product delivery.
Design, train, and evaluate AI models for super resolution, frame generation, neural rendering, and related real-time graphics workloads.
Develop training methods, loss functions, and evaluation approaches that improve visual quality, temporal stability, robustness, and model efficiency.
Define dataset requirements and work with data and infrastructure teams to improve training coverage, reference quality, and data pipelines.
Analyze visual quality and model failure cases, and drive improvements across models, training data, model inputs, and product integration.
Optimize AI models for real-time execution on Intel GPUs, balancing visual quality, latency, memory and compute costs.
Collaborate with SW, HW architecture, validation, and ecosystem teams to integrate, validate, and deliver product-ready technologies.
Evaluate emerging AI architectures and techniques, and help define the technical direction for future XeSS and AI graphics capabilities.
Qualifications: