Why pre-screen CFD analysts before the technical panel and modelling exercise
Pre-screening CFD analysts saves the technical panel from candidates who can only drive a GUI. Applicants arrive from mechanical and aerospace masters programmes, from vendor training in Fluent, STAR-CCM+ or OpenFOAM, and from consultancies where they ran templated cases. A resume lists solvers and y+ jargon but not whether they chose the turbulence model or inherited it, whether anyone tested the design afterwards, or what they did when residuals stalled. Ten minutes of spoken answers separates those three groups fast.
What actually matters when screening Computational Fluid Dynamics (CFD) Analyst candidates
- 01
Technical depth
Probe turbulence modelling, meshing, boundary conditions, and whether they understand the physics or only the solver.
- 02
Work that shipped
Look for simulations that drove a real design decision and were later checked against test or field data.
- 03
Diagnosis under uncertainty
Test what they do when a solution will not converge or the answer looks physically implausible.
- 04
Working across the org
Check how they present results to designers who will act on a colourful contour plot without reading the caveats.
Pre-screening questions to ask Computational Fluid Dynamics (CFD) Analyst candidates
12 questions grouped by what they test. Ask the same set in every screen and score answers on a consistent scale, or send them as an async video screen and compare answers side by side.
Physics and methods
Explain, on a basic level, how computational fluid dynamics works, as if you were talking to a design engineer with no CFD background.
Plain-language coverage of discretising a domain, solving conservation equations cell by cell, and iterating to convergence, without hiding behind jargon.
They recite equations or software menus and cannot explain what the solver is actually approximating.
Talk me through your experience with turbulence models: which ones you use most, and why you pick one over another.
Named models (k-omega SST, k-epsilon, Spalart-Allmaras, LES) tied to specific flow features like separation, wall treatment, y+ targets, and compute budget.
They always use the default model and cannot say what it does poorly.
Which discretisation methods have you worked with: finite volume, finite element, or boundary element?
Clear grasp of finite volume as the industry norm for flow, with a reason it suits conservation, plus any FEM work on FSI or thermal coupling.
They treat the three as interchangeable labels with no idea which their solver uses.
How deep does your background go in heat transfer and thermodynamics, and where has that mattered in a CFD job?
Concrete work on conjugate heat transfer, radiation models, buoyancy-driven flow, or electronics cooling, with the physics assumption they had to justify.
Coursework only, with no simulation where thermal physics changed the setup or the answer.
Tools and scale
Which CFD software do you use, and which one would you call yourself a specialist in?
A primary tool named with version-level familiarity (Fluent, STAR-CCM+, OpenFOAM, CFX) plus meshing tools such as Pointwise, ICEM, or snappyHexMesh.
A long list of packages with no depth in any, or exposure limited to a university lab licence.
What is your experience running CFD on high-performance computing clusters?
Job submission via SLURM or PBS, core-count scaling decisions, partitioning, wall-clock versus cell-count tradeoffs, and handling multi-gigabyte result files.
All work done on a single workstation with no sense of how runs scale or cost.
Have you developed or modified CFD code, for example custom UDFs, boundary conditions, or OpenFOAM solvers?
A named modification: a user-defined function in Fluent, a custom source term, a modified solver in OpenFOAM, and the reason the stock tool fell short.
They claim coding ability but cannot name a language, a file, or a problem it solved.
Work that shipped
Describe a time your CFD simulation changed a design. What did you recommend and what did the team do with it?
A specific geometry change, the metric that moved (drag count, pressure drop, peak wall temperature), and evidence the design team acted on it.
Simulations that produced reports nobody used, or they cannot say what changed in the hardware.
Walk me through how you validate and verify your CFD results.
Grid independence study, residual and imbalance checks, then comparison against wind tunnel, PIV, thermocouple, or field data with the actual delta quoted.
Validation described as checking that residuals dropped, with no external measurement ever involved.
Tell me about a run that would not converge or gave a physically implausible answer. How did you diagnose it?
A structured hunt: mesh quality metrics, boundary condition review, under-relaxation and courant number changes, initialisation, then checking the physics assumption itself.
They only lowered relaxation factors or ran longer until numbers stopped moving, calling that converged.
How do you present CFD results to designers who will act on a contour plot without reading your caveats?
Deliberate habits: leading with the design decision, quoting uncertainty ranges, choosing scales carefully, and flagging where the model is unreliable up front.
They blame designers for misreading plots and take no responsibility for how results are framed.
Sector fit and logistics
Which sectors have you done CFD in: aerospace, automotive, turbomachinery, process equipment, building services? And what does your availability look like for a technical panel?
Sector-specific detail such as certification-driven aero work, underhood thermal management, or ATEX process constraints, plus clear notice period and interview availability.
Sector claims that fall apart when asked what the typical Reynolds number or key metric was.
How to score responses
Score every candidate on the same four criteria immediately after the screen. At this stage you are shortlisting for panel interviews, not making the final call.
| Criterion | What a 5 looks like | Scale |
|---|---|---|
| Technical depth | Explains turbulence and boundary-condition choices from the physics, not from solver defaults. | 1 · 2 · 3 · 4 · 5 |
| Work that shipped | Names simulations that changed a design and were validated against test data, including the error margin. | 1 · 2 · 3 · 4 · 5 |
| Diagnosis under uncertainty | Treats non-convergence and implausible results as evidence, and isolates mesh, model, or setup as the cause. | 1 · 2 · 3 · 4 · 5 |
| Working across the org | Presents results with uncertainty attached so designers act on the finding rather than the picture. | 1 · 2 · 3 · 4 · 5 |
Async video lets you hear a CFD analyst explain turbulence model choice and a stalled residual plot in their own words, which is exactly how they will brief a designer who never reads the caveats.
Try it on HirevireScreening FAQ
Process basics
What should a CFD analyst screening cover before the technical panel?
Cover four areas: solver and meshing experience with named tools (Fluent, STAR-CCM+, OpenFOAM, CFX, Pointwise, Fidelity), the physics they can explain without a screen in front of them, one project where a simulation changed a design, and how they reported uncertainty. Leave detailed governing equations and code modification depth to the panel.
Do CFD analysts need a PhD or specific certifications?
No. A PhD matters for research-heavin roles involving solver development, LES, or novel physics; most industrial CFD work does not require one. Look instead for a mechanical, aerospace, or chemical engineering degree, hands-on meshing hours, and vendor training such as ANSYS or Siemens courses. Published validation work counts more than any certificate.
Evaluating answers
How do you tell a real CFD analyst from someone who only ran templated cases?
Ask why they picked a specific turbulence model and what they would lose with a different one. Real analysts talk about wall treatment, y+ targets, separation prediction, and where k-epsilon fails against k-omega SST. Template runners describe menu paths, default settings, and results without mentioning mesh independence or boundary condition sensitivity.
What does a strong validation answer sound like?
A strong answer names the comparison data: wind tunnel runs, pressure taps, PIV, thermocouple readings, or field measurements, plus the percentage delta and what explained it. They should mention grid convergence studies, sensitivity to inlet conditions, and a case where the simulation was wrong. Vague claims of results matching test data are not enough.
Go deeper on this role
Sanat has personally screened 300+ candidates across 18+ years of building and hiring teams, and turned that screening process into Hirevire's video screening platform. LinkedIn
