Member of Technical Staff - Foundation Model Architecture & AI Infrastructure

Role overview A senior technical role owning the architecture and scaling of a foundation model infrastructure layer that powers physics-based hardware simulation across Tier-1 semiconductor and electronics programs. The position focuses on evolving a unified operator foundation model, scaling distributed training, and designing trillion-voxel inference systems that run reliably in production.

Responsibilities - Design and refine transformer variants for structured spatial domains, including sparse and locality-aware attention mechanisms, hierarchical multi-resolution attention, and graph-transformer systems for multi-entity interactions. - Scale distributed training beyond current 45TB-scale datasets, improve generalization across heterogeneous operator distributions, design scalable data and curriculum strategies, and maintain reproducibility and determinism. - Architect trillion-voxel inference systems using sparse and hierarchical computation, balancing memory, compute, and communication while maintaining production-grade stability and determinism. - Ship expanded operator capabilities into production, increase simulation throughput substantially, support global multi-entity deployment, and maintain robustness under diverse industrial workloads. - Build feedback loops from deployed production environments into continuous learning systems.

Requirements - Deep experience in large-scale foundation model architecture and transformer variants (sparse, hierarchical, graph-based). - Strong background in distributed training systems and production ML system design. - Experience scaling structured datasets and modern ML stacks such as PyTorch and distributed training ecosystems. - Strong software engineering fundamentals with clean abstractions and scalable code design. - Strong CI, regression testing, and validation discipline. - Track record of shipping AI systems that run in production, not just research experiments.

Nice to have - Familiarity with operator regimes spanning Maxwell's equations, elasticity, plasticity, Navier-Stokes, nonlinear constitutive systems, and coupled multiphysics interactions. - Experience designing systems that function as durable infrastructure rather than short-lived prototypes.

Benefits and work setup - Full-time remote or hybrid position. - High-ownership role at a Series A-stage company defining a foundational abstraction layer early. - Equity participation and direct production impact on real Tier-1 hardware programs.