Principal Engineer

Location: Palo Alto, United States; hybrid

Salary: $100K–$220K annually

Company

hirify.global builds AI-enabled infrastructure and manufacturing-resolution physics simulation software for hardware development workflows.

What you will do

• Architect and deliver production-grade solver infrastructure capable of scaling beyond billions of degrees of freedom toward trillion-DOF systems.
• Develop additive and multiplicative Schwarz methods, overlapping and non-overlapping domain decomposition, and scalable coarse-space hierarchies.
• Implement algebraic and geometric multigrid, block- and physics-aware preconditioners, ILU variants, and sparse approximate inverses.
• Build CG, GMRES/FGMRES, BiCGStab, pipelined, communication-reducing, and mixed-precision solver methods.
• Design deterministic parallel reductions and distributed execution across multi-GPU clusters.
• Integrate AI-augmented components for coarse-space discovery, adaptive preconditioner selection, spectral approximation, and operator compression.

Requirements

• Deep expertise in domain decomposition, Schwarz methods, multilevel solvers, and scalable preconditioning.
• Experience with large sparse systems, numerical stability, conditioning, and parallel numerical algorithms.
• GPU-accelerated sparse linear algebra experience with CUDA; HIP is appreciated.
• Experience with kernel-level performance engineering, multi-GPU scaling, and distributed execution.
• Track record of shipping solver infrastructure into production rather than building prototypes only.
• Strong CI, regression, correctness-validation, and deterministic execution practices.

Nice to have

• Experience with AI-augmented numerical solver components.
• Experience across indefinite systems, saddle-point systems, strong coefficient jumps, anisotropy, or tightly coupled multiphysics blocks.

Culture & Benefits

• Small, technically focused startup team with deep domain expertise.
• High ownership and equity participation.
• Production impact across validated Tier-1 hardware ecosystem participants.
• Physics-first software based on verified methods rather than heuristics.