C++/Python AI Framework Engineer

C++ / Python Software Engineer - AI Frameworks (PyTorch)

Contract: B2B, 12 months, with extension possible · Work mode: remote, with occasional on-site visits in Gdańsk ·

About the project

Our client is a global technology leader whose chips run in data centres, laptops and edge devices around the world. Its engineering team in Gdańsk is making PyTorch run correctly and at full speed on the company's own data-centre AI accelerators. Much of the work goes upstream to open source, so your code reaches the global PyTorch community.

What you will work on

• Reproducing and root-causing framework bugs, then fixing them in C++ or Python
• Fixing and extending tests in the PyTorch test framework and keeping CI green
• Implementing operators, from simple ones up to complex, performance-critical paths
• Debugging and profiling on accelerator hardware to close performance and accuracy gaps
• Taking tasks end to end, from first reproduction to a merged upstream pull request
• Tracking down regressions with git bisect and benchmark suites

Task size grows with experience: junior engineers start with bug reproduction, test fixes and simple operators with guidance. More experienced engineers own larger features independently, including the upstream PR.

Must-have skills

• Prior hands-on development experience with the Gaudi AI accelerator software stack (Gaudi SW stack / Gaudi PyTorch bridge). Applications without Gaudi development experience cannot be considered.
• Strong C++ and Python
• Python testing with pytest and the PyTorch test framework (OpInfo, common_device_type / instantiate_device_type_tests, common_utils)
• Experience with PyTorch internals or the internals of another deep learning framework
• GPU/accelerator programming basics (SYCL, CUDA or OpenCL)
• Debugging and profiling skills
• CI with GitHub Actions and Jenkins
• Linux, Docker, git, including git bisect
• An understanding of deep learning and PyTorch fundamentals
• Familiarity with dynamo benchmark suites

Why join

• Deep, hands-on framework work on next-generation AI hardware, not glue code
• Upstream open-source contributions you can show in your portfolio
• A remote-first setup, with on-site visits only when a task needs them
• A stable 12-month B2B engagement
• A hiring process designed to move quickly, with a clear answer at each stage

How to apply

Send your CV and, if you have one, a link to your GitHub or upstream contributions. Please say briefly where and how you have worked with the Gaudi software stack.

You will be engaged through ALOIS Solutions, a global IT staffing partner operating in 21+ countries. We handle the contract, invoicing and onboarding, so you can focus on the engineering.