Develop the path from AI model architecture to physical silicon by creating training techniques and optimization strategies for novel compute substrates. You will focus on energy benchmarking, hardware-aware training, and low-level GPU kernel development to maximize efficiency.
Develop and optimize GPU-accelerated simulation frameworks to test physics-based computing systems for machine learning. Collaborate with hardware and algorithm teams to integrate device models and ensure reproducible experiment tracking.
Develop foundational language and reasoning models tailored for unconventional silicon to achieve extreme energy efficiency. Collaborate with hardware designers to co-design model architectures that leverage the physical dynamics of semiconductors.
Define and implement comprehensive verification plans and environments for IP blocks and full chips from specification to tape-out. Collaborate with cross-functional teams to debug complex hardware/software interactions and ensure 100% verification closure.
Drive the invention, prototyping, and validation of core components for a novel, energy-efficient computing platform. Work across theoretical modeling, algorithm development, and hardware/software co-design to map neural networks to device physics.
Develop physics-based system models and GPU-accelerated simulations to evaluate unconventional computing systems for ML workloads. Collaborate across teams to optimize the trade-offs between algorithms and hardware through high-fidelity differential equation solvers.
Develop high-fidelity power, performance, and area estimation tools for novel AI acceleration architectures. Collaborate across teams to create comparative analyses and support high-level system design and hardware verification.
Develop high-performance PyTorch components to model complex, time-varying dynamic systems for next-generation AI architectures. Build simulation frameworks that enable rapid iteration across physics-based computing systems for machine learning workloads.
