Founding Physical AI Engineer (AI Design Engine)
Job Description
We are looking for the first team member to build an AI design tool that shrinks the heat exchanger invention loop from months to days.
\n
\n
About SwirlX
\n
SwirlX, Inc. is a Delaware C Corp headquartered in La Jolla, California. We build high performance liquid to liquid heat exchangers using metal 3D printing (LPBF). Our products target liquid cooling for AI data centers.
\n
\n
Why This Role
\n
- \n
- A true closed loop environment where design (geometry), simulation (CFD), manufacturing (LPBF), and thermal test data are all generated inside the company
- Designs proposed by your model can be printed and physically tested within weeks
- As our first AI hire, you own the architecture and technology stack decisions
- Direct exposure to US data center customers and their real requirements
\n
\n
\n
\n
\n
\n
What You Will Do
\n
- \n
- Build a design engine that takes requirements (heat load, pressure drop, envelope) and outputs manufacturable geometry with predicted performance
- Structure geometry, simulation, process, and test data into machine readable, version controlled training assets
- Develop surrogate performance models that embed physics priors (conservation laws, dimensionless numbers, heat transfer correlations)
- Quantify and close the sim to real gap between synthetic simulation data and measured data
- Build an active learning loop in which the model proposes the next experiment and the next design
- Implement automated workflows where LLM agents directly call CAD, meshing, and CFD tools
\n
\n
\n
\n
\n
\n
\n
\n
Success Metrics
\n
- \n
- Within 6 months, cut the time for one design iteration to one tenth of today's baseline
- Within 12 months, demonstrate that an AI proposed design outperforms our existing design in physical testing
\n
\n
\n
\n
Who We Are Looking For (evidence of two or more)
\n
- \n
- You have built engineering simulation or CAD software, or automated simulation workflows in code
- You hold an engineering or physics degree (mechanical, materials, chemical, computational, or similar) and have shipped real ML projects
- A model you trained has driven real product, process, or experimental decisions
- Papers, open source work, or a portfolio in 3D geometry learning (GNNs, neural operators, implicit representations)
- You have designed or shipped real hardware
\n
\n
\n
\n
\n
\n
\n
Required Skills
\n
- \n
- Able to implement models in PyTorch and run experiments end to end on your own
- Able to diagnose data quality issues using physical intuition
- Able to debate design tradeoffs with thermal and fluids engineers in their
- Comfortable collaborating across time zones with our Seoul manufacturing team, including occasional travel to Korea
- Authorized to work in the United States
\n
\n
\n
\n
\n
\n
\n
Nice to Have
\n
- \n
- CFD or FEA tool automation experience (Ansys, OpenFOAM, STAR CCM+)
- Manufacturing domain data experience in additive manufacturing, semiconductors, or aerospace
- Experience building LLM agents and tool calling systems
- Experience running cloud GPU training environments and MLOps
- Korean ability
\n
\n
\n
\n
\n
\n
\n
Compensation and Benefits
\n
- \n
- Base salary range: $80,000 to $200,000, depending on demonstrated ability
- Stock options
- Health benefits and unlimited paid time off
- No minimum years of experience required
\n
\n
\n
\n
\n
\n
How to Apply
\n
- \n
- Instead of a resume, send us one page on the single thing you built that you are most proud of. Focus on the problem, the data, the model, and how the result was physically validated.
\n