Developing Olympiad-level physical intelligence in robots to enable them to anticipate and reason about future movements of any object or scene at the level of a race car driver or professional athlete
Requirements
- Proven experience in scaling and optimizing large AI models, with a strong understanding of performance-related codesign
- Proficiency in Python and a deep understanding of software engineering best practices
- In-depth knowledge of deep learning fundamentals, including optimization techniques, loss functions, and neural network architectures
- Experience with deep learning frameworks such as PyTorch, TensorFlow, or JAX
- Practical experience leveraging GPUs, SIMD instructions, multithreading, or custom accelerators (e.g., TPUs, edge NPUs) for AI model inference and optimization
- Deep understanding with bottlenecks of inference hardware – compute throughput, memory bandwidth, and interconnect
Responsibilities
- Design, train, and evaluate models optimized for edge accelerators
- Focus on improving model quality and training stability with model architecture design
- Conduct scaling laws for model architecture and parallelism-aware compute efficiency
- Design novel model architectures and algorithms for scaling and hardware utilization
- Innovate in domains such as sparsity, distillation, quantization and parallelism
- Profile inference performance to ensure model architecture maximizes hardware efficiency
Other
- Demonstrated ability to work collaboratively in a cross-functional team environment
- Strong problem-solving skills and the ability to troubleshoot complex system-level issues
