ML Performance Architect

ML Performance Architect

I am seeking a ML Performance Architect to join my start-up semiconductor client in London developing a unique new processor technology, all software and hardware from scratch, to run the world’s largest language models and optimise speed. The position requires a breadth of experience working on performance architecture of NPU / GPU / CPU / AI accelerators and at a senior level, as you’ll be operating on your own prior to the hiring of additional architects.

Primary responsibilities

• Develop new architectural features to help design next generation hardware
• Build a simulation of network hierarchies, to put together multi-core architectures
• Evaluate performance of cutting-edge AI workloads, characterisation and mapping
• Design space exploration

• MSc or PhD with 5+ years industry experience working on performance architecture of NPU / GPU / CPU / AI accelerators
• Software-hardware co-design
• work
• Deep learning frameworks, PyTorch, TensorFlow
• C/C++, Python

• MLOps, including model training, quantisiation, sparsity, model preprocessing etc.
• ML Compiler stack
• Verilog / RTL designs

£80-120,000 (depending on experience and technical match) amongst other benefits

Hybrid working

Interested? This is a great opportunity for a MSc or PhD-educated ML Performance Architect. Please apply now for immediate consideration and speak with Chris Wyatt  who is recruiting for this position in London, Greater London, UK.