About the role
About Voltai
Voltai is developing world models, and agents to learn, evaluate, plan, experiment, and interact with the physical world. We are starting out with understanding and building hardware; electronics systems and semiconductors where AI can design and create beyond human cognitive limits.
About the Team
Backed by Silicon Valley’s top investors, Stanford University, and CEOs/Presidents of Google, AMD, Broadcom, Marvell, etc. We are a team of previous Stanford professors, SAIL researchers, Olympiad medalists (IPhO, IOI, etc.), CTOs of Synopsys & GlobalFoundries, Head of Sales & CRO of Cadence, former US Secretary of Defense, National Security Advisor, and Senior Foreign-Policy Advisor to four US presidents.
What You'll Work On
- Develop and scale MPI+CUDA PDE solvers for electrostatics, charge transport, and electromagnetic field problems on complex 3D IC geometries across multi-node GPU clusters
- Tune and extend AMG preconditioners, Krylov solvers, and mesh pipelines for performance and correctness at scale
- Build and train neural operators (FNO, DeepONet, GNO, and variants) as high-fidelity surrogates for PDE-based field solvers
- Design simulation pipelines that generate training data for neural operator models — including sampling strategies, mesh handling, and physical consistency checks
- Validate everything: analytical solutions, published benchmarks, and cross-validation between field solvers and learned surrogates
Required
- PhD in computational physics, applied mathematics, computational engineering, or a closely related field
- Deep expertise in numerical PDE methods: FEM, FVM, or BEM — weak formulations, quadrature, convergence, error analysis
- Strong C++ and CUDA — writing and optimizing kernels, memory hierarchy, multi-GPU programming
- Multi-node HPC: MPI, domain decomposition, collective communication, strong/weak scaling
- Sparse linear algebra at depth: Krylov methods, algebraic multigrid, preconditioning strategies
- Hands-on experience with neural operators (FNO, DeepONet, or equivalent) — training, architecture design, and evaluation on PDE datasets
- Solid understanding of AI for Science methodology: how to design datasets from simulations, handle out-of-distribution generalization, and ensure physical consistency of learned models
Strongly Preferred
- Experience with HYPRE, PETSc, and Trilinos
- Familiarity with multi-node GPU clusters: NCCL, CUDA-aware MPI, NVLink topologies
- Published work in neural operators, physics-informed ML, or scientific HPC
- IC design domain knowledge: device physics, semiconductor materials, layout data formats