ML Training Systems Engineer - Performance Focus

Internetwork Expert; Amsterdam; Hybrid; EUR 100,000 - 150,000; Full time

Benefits offered by this job

High base salary

Generous bonus structure

Cutting-edge hardware and software

Job summary

Internetwork Expert is seeking an ML Engineer to speed up large-scale model training by optimizing the internal stack and compute infrastructure. You will work across the full training pipeline—from GPU kernels to system-level throughput—using profiling, CUDA-level tuning, and distributed systems techniques.

This role focuses on reducing training time, boosting iteration speed, and improving compute efficiency within a growing ML training systems team in Amsterdam.

Qualifications

• Experience optimizing neural network training in large-scale production or research settings.
• Hands-on experience with PyTorch or JAX.
• Experience with CUDA, Triton, or other low-level GPU technologies for performance tuning.
• Strong Python for infrastructure tooling and integration with ML frameworks.

Responsibilities

• Optimize the model training pipeline for speed and reliability across the full stack.
• Apply GPU-level optimizations to improve training performance at scale.
• Identify and resolve performance bottlenecks from data loading to CUDA kernels.
• Build tools and extend internal infrastructure for scalable, reproducible training workflows.
• Mentor engineers and researchers in performance best practices.
• Grow the team’s GPU and systems-level capabilities and drive fast experimentation.

Skills

Python

PyTorch/JAX

Profiling/Debugging

Distributed training concepts

Tools

Nsight

CUDA profiling tools

Torch profiler

Job description

Internetwork Expert is seeking an ML Engineer to speed up large-scale model training by optimizing the internal stack and compute infrastructure. You will work across the full training pipeline—from GPU kernels to system-level throughput—using profiling, CUDA-level tuning, and distributed systems techniques.

This role focuses on reducing training time, boosting iteration speed, and improving compute efficiency within a growing ML training systems team in Amsterdam.