ML Optimisation Engineer

About The Position

We are looking for an ML Optimisation Engineer to join our Research and Development department at Squarehead Technology. A complete and effective infrastructure is already in place to train machine learning models, but we need to expand our capabilities to speed up and improve training and inference. While our platform engineers build and maintain the systems that power ML workflows, this position focuses on optimisation, making models run faster and more efficiently.

As an ML Optimisation Engineer, you will enable us to support larger and more advanced models, such as spatial audio models, by reducing their computational demands through techniques such as quantisation, pruning, distillation, and low-level tuning, and you will interface with the Software team to integrate the models in the final product.

You will work closely with a team of scientists with expertise in signal processing, array technology, audio event detection and tracking, classification, and data science. The position offers the opportunity to contribute in a collaborative environment where machine learning and signal processing come together to solve real-world engineering problems.

Responsibilities

• Model compression & quantisation: Implement post-training quantisation (PTQ), quantisation-aware training (QAT), pruning, and knowledge distillation to shrink models while maintaining accuracy.
• Hardware-specific optimisation: Integrate and fine-tune models on runtimes such as TensorRT, ONNX Runtime, or CoreML; leverage CUDA/cuDNN for acceleration.
• Profiling & benchmarking: Use tools such as Nsight, TensorBoard Profiler, or custom benchmarks to identify compute and memory bottlenecks.
• Audio & signal-processing pipelines: Optimise DSP algorithms (resampling, FFT, filter banks) for performance and robustness.
• End-to-end model integration: Collaborate with ML Scientists during the research phase and take ownership of implementing, optimising, and deploying models into production systems.
  
  

About You

Personality & Interests

We value curiosity, teamwork, and the ability to work independently when needed. We are looking for someone with an interest in machine learning as an engineering challenge, with strong coding skills and a motivation to optimise systems for speed, robustness, and reliability.

Core Competencies

• Deep learning internals: Proficiency in PyTorch, TensorFlow, or JAX, with experience in graph editing and custom operator development.
• Low-level programming: Skills in C++ and GPU acceleration framework (CUDA/Vulkan Compute Shaders) for kernel development, mixed-precision computation, and memory-footprint optimisation.
• DSP & audio fundamentals: Knowledge of digital signal processing theory and implementation.
• Performance mindset: Ability to optimise throughput under resource constraints and set up regression tests to guard against performance or accuracy regressions.
  
  

Workplace & Benefits

• Office location in Nydalen, Oslo, with easy access to the subway
• Flexibility on partly remote work by agreement
• Staff canteen with subsidised lunch
• Organised social and sports activities (Paddel-Tennis in 2025)
  
  

Practical

We review applications and invite candidates for interviews on an ongoing basis, so if you are interested in the position, we recommend that you apply quickly.

We work with partners who require confidentiality and have security requirements that we must ensure our employees can comply with. This position may require security clearance.