Want to know if this job is worth applying to?
Palo Alto, California, United States
Hybrid
We are building an AI assistant for hardware design. Our product helps engineers explore, analyze, and optimize hardware systems, integrating cutting-edge AI with the realities of engineering design workflows.
Hands-on technical role spanning validation, training data, and customers. Vinci's physics foundation model produces solver-accurate thermal results on full-resolution hardware — real geometry, nanometer-scale features, no simplification — across conduction, convection, and thermoelasticity. You'll prove that case by case: validating new solvers and methods against competitor FEA tools and measured data, generating the reference cases that train the model, running customer test cases end to end, and bringing what you learn back to research and product as feature definition. The ideal candidate has a strong thermal simulation foundation, reasons from first principles, and is comfortable presenting a result to a customer's thermal engineers who will not take your word for it.
Validate new solvers and methods: build test cases that stress steady-state and transient conduction, convection, and thermoelastic stress and warpage, and benchmark them against competitor FEA tools and measured data.
Help set the accuracy bar for release, and say so when it isn't met.
Generate training data: build and run the reference cases that feed model training — geometry sets, material stacks, boundary conditions, parameter sweeps — and curate the results the model learns from.
Serve as the product's heaviest internal user: find the bugs, bad defaults, and workflows that break on a real 2.5D package or a 12-layer board, and file them with a root cause attached rather than a screenshot.
Run customer cases end to end: take a customer's native geometry and power maps, produce a result, interpret it, and present it.
Support customer onboarding and training, and own the technical questions that follow.
Build your own tooling: test harnesses, batch sweeps, regression comparisons, and post-processing, prototyping quickly with AI coding assistants.
Turn temperature and displacement fields into plots, side-by-side design comparisons, and decks that hold up in front of a customer's engineering leadership.
Bring field observations back to research and product as specific requests: what to build, what the output needs to show, which physics matters to which customer.
BS or higher in mechanical engineering, electrical engineering, aerospace engineering, or a comparably rigorous technical field.
3+ years in thermal or simulation engineering; graduate research counts.
Hands-on thermal or thermo-mechanical CFD/FEA in Ansys Icepak, Flotherm, Fluent, Ansys Mechanical, COMSOL, or equivalent, including a working sense of where each tool's results degrade.
Practical meshing knowledge: mesh strategy, element selection, where and how to refine (interfaces, thin layers, boundary layers, high-gradient regions), and how to qualify a mesh through independence studies, convergence behavior, and quality metrics such as aspect ratio and skewness. Our meshing is automated; evaluating what it produces takes the same judgment.
Ability to reason from first principles: when a result looks wrong, to determine whether the cause is the physics, the boundary conditions, the geometry, or the software.
CAD fluency: opening, navigating, interrogating, and repairing real geometry in mechanical CAD, ECAD, or package layout.
Strong communication skills, including the ability to explain a result and its limits to a principal thermal engineer and to a VP in the same meeting, and to build the plots and slides yourself.
Coding experience. Python is what we use, though any real scripting background counts (MATLAB, LabVIEW, C++, OpenFOAM customization). What matters is that you write your own scripts, internal tools, and one-off analysis, including with AI coding assistants.
10+ years of thermal engineering, with shipped hardware behind it.
Semiconductor packaging or high power-density electronics: chip–package–board thermal, 2.5D/3D stacks, chiplets, power maps, TIMs, warpage and reliability, cold plates, vapor chambers, liquid cooling.
Customer-facing experience in applications, solutions, or field engineering, including demos, trainings, and presentations you delivered yourself.
Experience validating or building a solver rather than only running one.
Comfort working in Linux: command line, GPU/HPC job submission, version control.
You'll work on semiconductor and advanced packaging hardware first — AI accelerators, chiplet assemblies, high-power boards — then robotics, electric vehicles, power electronics, aerospace, and defense as those customers come on.
You'll work directly with the researchers and GPU kernel engineers building the model, not through a support tier.
Customers use these results for production sign-off, so validation calls carry real consequence.
We're in Palo Alto, CA near the Caltrain station, working 2–3 days a week in person. We hire remote for the right person — members of the team already work this way. Travel is occasional.
Reports to the Founding Mechanical Engineer, who reports to the CTO.