Senior Applied Research Engineer

Senior Applied Research Engineer | Barcelona, Spain

We’re partnering with a high-growth AI company building cutting-edge foundation models designed to solve complex enterprise decision-making problems.

This is an opportunity to join a deeply technical team working on large-scale model training, efficiency, and productionisation. The environment is highly research-driven, but with a strong focus on real-world performance, scalability, and impact.

The role

You’ll work on the performance and efficiency of large-scale training workloads, helping to improve how advanced models are trained, scaled, optimised, and served in production.

This role sits at the intersection of research and systems engineering, with a strong focus on distributed training, profiling, memory optimisation, and model efficiency.

This is the chance to work on genuinely hard problems in foundation model development with meaningful real-world application.

What you’ll be doing

• Profile end-to-end distributed training runs to identify bottlenecks across compute, GPU memory, and inter-GPU communication
• Improve the efficiency and reliability of large-scale training jobs, including contributing to architectural decisions and developing Triton/CUDA kernels where needed
• Design and implement model scaling, parallelisation, and memory optimisation techniques for very large context training workloads
• Partner closely with ML Researchers to diagnose inefficiencies, ensure new ideas scale effectively, and share best practice around model performance
• Support the productionisation and serving of models from the research side, including improving inference efficiency through techniques such as quantisation

📍 Barcelona, Spain - hybrid working [Firm will offer relocation support for Barcelona]

💶 Highly competitive salary + benefits + equity

⏱️ Permanent Role

What you bring:

• Strong understanding of modern ML architectures and large-scale training pipelines
• Experience running distributed training jobs across multi-GPU systems
• Advanced profiling and debugging skills across CPU, GPU, memory, latency, and inter-GPU communication
• Strong Python skills
• Experience with model scaling and parallelisation approaches, including tensor and pipeline parallelism
• Familiarity with NCCL, MPI, and distributed communication primitives - highly desirable
• Knowledge of PyTorch and Triton internals - highly desirable
• Experience with C++ and CUDA - highly desirable

If you are interested in this role, please respond directly to this advert with your updated CV or email it to chantelle.smith@sourcegroupinternational.com