Research Software Engineer - Scientific Computing & GPU Infrastructure

LinuxCareers; München; On-site; EUR 70,000 - 110,000; Full time

Benefits offered by this job

Vacation allowance

Public transport subsidy

Health insurance

Learning budget

Relocation support

Pension plan

Modern Munich office

Team events

Phone bill reimbursement

Wellness program

Job summary

Marvel Fusion in Munich seeks a Research Software Engineer to develop scientific Python software and GPU infrastructure for fusion research. You will collaborate with scientists, verify numerical results, profile performance, and deploy workloads on NVIDIA GPU servers or AWS with reproducible environments.

The ideal candidate has a strong Python background, experience in scientific computing, GPU frameworks (CuPy/PyTorch/JAX/CUDA), and a structured approach to debugging, testing, and

Qualifications

• Bachelor's degree or equivalent practical experience in a technical field.
• Strong Python software engineering skills with maintainable code.
• Experience in scientific or numerical computing and testing accuracy.
• Experience profiling workloads and GPU frameworks (CuPy, PyTorch, JAX, CUDA).
• Experience deploying Python/GPU workloads on Linux servers or cloud with reproducible envs.

Responsibilities

• Develop and maintain scientific Python software with reusable, tested features.
• Collaborate with scientists to verify results against reference cases and accuracy requirements.
• Profile workloads and compare algorithms across CPU/GPU implementations with benchmarks.
• Deploy workloads on NVIDIA GPU servers or AWS with reproducible environments.
• Design APIs for job submission, status tracking, and result retrieval.
• Improve testing, CI, packaging, and documentation; deliver development tasks.

Skills

Python

GPU computing

Profiling

Version control

Automated testing

Documentation

English communication

Education

Bachelor's degree in Computer Science, Physics, Applied Mathematics, Engineering, or related field

Equivalent practical experience

Tools

CuPy

PyTorch

JAX

CUDA

AWS GPU services

Job description

Marvel Fusion is Europe's leading fusion energy company seeking a Research Software Engineer to develop scientific computing and GPU infrastructure. You will combine software expertise with physics and engineering knowledge to create reliable scientific software that advances computational challenges in fusion research.

What You'll Do

• Develop and maintain scientific Python software with reusable, tested features and clear interfaces aligned with research requirements
• Collaborate with scientists to define acceptance criteria and verify numerical results against reference cases and accuracy requirements
• Use profiling to identify performance bottlenecks and compare algorithms, code translations, and CPU/GPU implementations through reproducible benchmarks
• Deploy scientific workloads on NVIDIA GPU servers or AWS using reproducible environments and automated execution workflows
• Design and maintain APIs for job submission, status tracking, and result retrieval to connect computational services to internal applications
• Improve testing, continuous integration, packaging, documentation, and independently deliver agreed development tasks

What You Need

• Bachelor's degree in Computer Science, Physics, Applied Mathematics, Engineering, or related field, or equivalent practical experience
• Strong Python software engineering skills with a track record of independently delivering maintainable software within existing codebases
• Experience with scientific or numerical computing, including testing numerical accuracy and collaborating with domain experts
• Practical experience profiling computational workloads and using GPU computing frameworks such as CuPy, PyTorch, JAX, or CUDA
• Hands-on experience deploying scientific Python or GPU workloads on remote Linux servers or cloud infrastructure with reproducible environments and data handling
• Structured approach to debugging, version control, automated testing, and documentation with ability to make pragmatic engineering decisions
• Self-driven, collaborative working style with excellent written and verbal communication skills in English

Nice to Have

• Direct experience with AWS GPU services or shared NVIDIA GPU infrastructure including containers, workload scheduling, or monitoring
• Expertise in CUDA, multi-GPU computing, high-performance computing, or performance-portable numerical software
• Experience integrating scientific applications with compute services, or background in optics, photonics, or experimental data analysis
• Experience evaluating AI-based acceleration of scientific workloads against accuracy and runtime benchmarks

30 days of vacation plus additional half day on 24th and 31st December, public transport subsidy, employee incentive program, phone bill reimbursement, Wellpass or complimentary private health insurance, professional learning and development budget, relocation support, company pension plan, modern office in central Munich, healthy office snacks, regular team events