Accelerated Physics Simulation Engineer – Agentic Computational Engineering (ACE)

Voyager Technologies, Inc.•Washington, DC

4d•$165,000 - $250,000

About The Position

We are seeking an Accelerated Physics Simulation Engineer. In this role, you will develop fast, high-fidelity physics simulation capabilities that allow AI agents to evaluate and optimize hardware designs over millions of design iterations. You will work at the intersection of numerical methods, GPU computing, and machine learning surrogates. You will help build differentiable and surrogate physics models that can be called directly by ACE agents, and you will validate them against high-fidelity solvers and real test data. This role is ideal for a computational scientist or engineer who loves PDEs, GPUs, and turning overnight runs into millisecond-scale kernels—and who uses AI tools as a force multiplier, not a curiosity. You will be joining the Agentic Computational Engineering (ACE) team, a specialized group within our Advanced Technology Development organization. ACE is responsible for building Voyager’s Generative Engine – the AI-native platform that compresses complex hardware development cycles from years to days. We design agentic AI systems that pair deeply with physics simulation, test data, and modern manufacturing so that Design for Manufacturing (DfM), Design for Assembly (DfA), and Design for Test (DfT) are built into the very first line of code and the very first sketch of a design.

Requirements

PhD in Computational Physics, Mechanical or Aerospace Engineering, Applied Mathematics, Computer Science (with a focus on numerical methods), or a related field; or Master’s degree + 3 years of highly relevant experience.
0–3 years of post-PhD industry, startup, or postdoctoral experience (or 3–6 years total experience working in computational science/engineering).
Hands-on experience implementing numerical methods for PDEs (e.g., FEM, FVM, FDM, particle or mesh-free methods) in research or production environments.
Experience with at least one major scientific computing or ML framework (e.g., JAX, PyTorch, TensorFlow) and one GPU or performance-oriented technology (e.g., CUDA, PhysicsNEMO, etc).
Demonstrated experience speeding up simulations or building surrogate models for physics problems, with quantitative before/after results.
Demonstrated “AI-first” workflow: you use LLMs to help generate, refactor, and test code so you can spend more time on modeling and physics.

Nice To Haves

Experience with CFD, structural mechanics, heat transfer, or plasma physics as applied to aerospace or propulsion systems.
Experience with electrical, power, and electromagnetic simulations as applied to PCB or RF systems.
Prior work on physics-informed neural networks (PINNs), neural operators (FNO, UNO, etc.), or other ML-based surrogates for physical systems.
Experience coupling commercial or open-source solvers (e.g., Ansys, COMSOL, Star-CCM+, OpenFOAM) with custom automation or optimization code.
Familiarity with differentiable programming and adjoint methods for design optimization.
A track record of side projects, open-source contributions, or competition results that demonstrate deep enthusiasm for computational physics and performance engineering.

Responsibilities

Design and implement fast physics solvers (e.g., CFD, thermal, structural, plasma) suitable for use inside agentic optimization loops.
Develop surrogate models (e.g., physics-informed neural networks, neural operators, graph neural networks) that approximate high-fidelity simulations at orders-of-magnitude lower cost.
Integrate accelerated solvers and surrogates into the ACE platform so AI agents can call them as tools during design and optimization.
Work with the ACE Applications Lead (Mechanical/Propulsion) to identify key regimes and quantities of interest and to ensure that accelerated models remain physically credible.
Create and curate training and validation datasets by coupling commercial or open-source solvers (e.g., Ansys, COMSOL, Star-CCM+, OpenFOAM) with automated parameter sweeps.
Profile and optimize GPU kernels and numerical pipelines, targeting large speedups over baseline codes while preserving required accuracy.
Develop test harnesses, benchmarks, and diagnostics that track accuracy, stability, and performance of accelerated models over time.
Use LLMs to accelerate boilerplate coding, experiment scripting, and documentation so you can focus on core numerical and physical insights.
Leverage the most advanced LLMs and tooling to assist with complex mathematics and numerical simulation generation.