GPU CUDA & AI Engineer — LLM Inference & HPC (Flexible Hours)

MWDN Ltd.; Barcelona; Remote; EUR 90,000 - 140,000; Full time

Benefits offered by this job

English courses

Flexible working hours

HR and legal support for contractors

Independent contractor support

Paid time off: 29 days plus holidays

Job summary

MWDN Ltd. is seeking a highly technical engineer or researcher with deep expertise in GPU computing, CUDA, and LLM inference optimization to work on challenging AI/HPC problems.

The role offers opportunities to work on international products within a stable environment and collaborations across multiple regions. The ideal candidate will have strong CUDA/GPU programming experience, in-depth knowledge of GPU architectures, and experience with LLM inference frameworks.

Qualifications

• Strong hands-on CUDA and GPU programming required.
• Deep GPU architecture, memory hierarchy, parallel execution knowledge.
• Experience with LLM inference and AI systems.
• English communication skills and problem solving.

Responsibilities

• Investigate and optimize LLM inference on modern GPUs.
• Work with vLLM, SGLang, TensorRT-LLM or similar frameworks.
• Analyze GPU execution, memory access, kernel performance.
• Develop, optimize, or reverse-engineer CUDA kernels.
• Explore optimizations for matmul, attention, and NN layers.
• Benchmark to improve latency, throughput, GPU utilization.
• Tackle challenging problems with hands-on experimentation.

Skills

CUDA programming

GPU architecture

LLM inference

English communication

Performance debugging

Education

PhD or research exp in CS/ML/Architecture

Tools

vLLM

SGLang

TensorRT-LLM

Triton

FlashAttention

PyTorch internals

Job description

MWDN Ltd. is seeking a highly technical engineer or researcher with deep expertise in GPU computing, CUDA, and LLM inference optimization to work on challenging AI/HPC problems.

The role offers opportunities to work on international products within a stable environment and collaborations across multiple regions. The ideal candidate will have strong CUDA/GPU programming experience, in-depth knowledge of GPU architectures, and experience with LLM inference frameworks.