> Markdown version of [/jobs/ext/3645942-lpa-design-engineer](https://www.wearedevelopers.com/jobs/ext/3645942-lpa-design-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# LPA Design Engineer

- **Company:** Muraytech Gmbh
- **Location:** Hamburg, Germany
- **Contract:** Permanent contract
- **Skills:** Python (Programming Language), Machine Learning, Network Architecture, PIC Microcontroller
- **Published:** October 9, 2026
- **Apply:** https://www.physicsworldjobs.com/external-redirect-registration?JobId=33645&LinkSource=JobDetails

## About the Role

* Physics background: deep understanding of laser-plasma acceleration, injection, dephasing, depletion, external and self-guiding physics.
 * PIC simulation expertise: hands-on with FBPIC, WarpX, or a comparable code.
 * Optimisation track record: applied use of Bayesian optimisation or similar methods for multi-objective search.
 * Machine learning skills: building surrogate models from simulation data. Physics-informed AI that encodes known physics into the network architecture.
 * Statistical judgment: for tolerance work, you characterise a robust basin, not just find a peak.
 * Software and HPC fluency: Python, reproducible simulation pipelines, cluster-scale sweeps.

## Description

As LPA Design Engineer, you'll own the working point design of MuRay's high-energy, high-charge laser-plasma accelerator. Concretely, you will:

 * Design density profiles that maximise high energy and minimise low energy charge. Real hardware sets the limits.
 * Extend the simulation chain from reduced-dimensionality models to full 3D. This captures effects a symmetric model misses.
 * Build an efficient optimisation loop that respects the cost of each simulation. Map the Pareto front between muon yield and radiation dose.
 * Develop a surrogate model as a digital twin of the LPA. Implement physics-informed architectures.
 * Run tolerance and jitter studies to test robustness of the chosen working point against real-world variation.