- Location
- SHANGHAI 07, China
- Type
- Full-time
- Department
- Engineering
- Seniority
- Lead
- Education
- PhD
- Source
- Workday
Description
At Cadence, we hire and develop leaders and innovators who want to make an impact on the world of technology.
## About the role
We build the finite-element engine behind our electrothermal and thermomechanical simulation products — the solver that our customers use to predict temperature, warpage, and stress in advanced packages, 3D-IC stacks, and PCBs. It is a large, production C++ codebase that meshes real industrial geometry with hundreds of millions of degrees of freedom and solves it in parallel.
You will join as a R&D engineer working directly with the team of simulation developers. This is a hands-on numerical-methods role: you will own features end to end — formulation on paper, implementation in C++, verification against analytical or reference solutions, and performance work on real customer cases.
## Responsibilities
- Derive, implement, and verify finite-element formulations in production C++.
- Write unit and regression tests, including verification against analytical
solutions and benchmark problems; treat a feature as unfinished until it is tested.
- Profile and optimize solver and meshing code for large models — both runtime and memory.
- Debug numerical issues on real customer cases: non-convergence, mesh quality, ill-conditioning, and accuracy loss.
- Write up formulations, assumptions, and validation results clearly enough that a colleague can reproduce them.
- Work with application engineers, QA, and product management to get features into customers' hands, and respond to what comes back.
## Required qualifications
- **PhD in Mechanical / Civil / Aerospace Engineering, Computational Mechanics,
Applied Mathematics, or a related field** (new graduates welcome), **or an MS in a
related field with 2+ years of relevant industry or research experience.**
- Solid grounding in **continuum mechanics and the finite element method**: weak
forms, isoparametric elements, numerical integration, assembly, and the
Newton–Raphson method for nonlinear problems.
- Working knowledge of **heat transfer and/or thermomechanical stress analysis** —
thermal eigenstrain, CTE mismatch, and the behaviour of layered structures.
- **Strong C++ (C++11 or later)**: you can read and safely extend a large existing
codebase, not just write greenfield code. Comfortable with pointers, memory
ownership, templates, and a debugger.
- Familiarity with **sparse linear algebra** and at least one sparse solver or
library.
- Ability to validate your own work: you reach for an analytical solution, a refined
mesh, or a patch test before you claim something is correct.
- Clear written and spoken English, and a genuine willingness to work as part of a
team — asking for review, giving it, and sharing what you learn.
## Preferred qualifications
- Experience with a commercial or open-source FEA code (Abaqus, ANSYS, LS-DYNA,
COMSOL, Code_Aster, FEniCS, deal.II, MOOSE) — as a developer, not only a user.
- Nonlinear FEA: large deformation, contact, plasticity, creep, or viscoplasticity.
- Parallel programming: OpenMP, MPI, or multithreaded C++.
- Direct and iterative sparse solvers (MUMPS, PARDISO, SuperLU, hypre/BoomerAMG,
PETSc) and graph partitioning (METIS/ParMETIS).
- Mesh generation, mesh quality metrics, or computational geometry.
- Semiconductor packaging or PCB background: 3D-IC, HBM stacks, solder reliability,
warpage.
- Development on Windows, and on Linux; Git or Perforce.
