Lead of Machine Learning Systems, Scale and Performance

Responsibilities

• Technical Leadership: Define the long-term technical roadmap and drive the development of scalable, high-performance ML systems.
• Algorithm Optimisation: Optimise state-of-the-art algorithms and architectures from the latest deep learning research for compute efficiency and performance.
• System Scaling: Design strategies for scaling machine learning models across diverse hardware platforms (GPU/TPU) and optimising system performance under heavy load.
• Low-Level Optimisation: Write efficient Python, C/C++, XLA, Pallas, Triton, or CUDA code to achieve performance breakthroughs.
• ML Systems Design: Architect robust distributed systems for training, deployment, and monitoring, ensuring computational efficiency and scalability.
• Data Pipeline Automation: Develop automated pipelines for data processing, model training, validation, and deployment, enabling efficient handling of large datasets.
• Team Collaboration: Partner with research, applied, and product teams to build a cohesive software stack supporting key projects.
• Mentorship: Guide and mentor the ML engineering team, fostering best practices in coding, testing, and documentation. 

Required Skills

• Expertise with Python and/or C/C++
• Understanding of Linux systems, performance analysis tools, and hardware optimisation techniques.
• Development with machine learning frameworks (JAX, Tensorflow, and/or PyTorch)
• Passion for profiling, identifying bottlenecks, and delivering efficient solutions.
• Fundamentals of modern Deep Learning

Desired Skills

• Track record of successfully scaling ML models.
• Experience writing custom CUDA kernels or XLA operations.
• Understanding of GPU/TPU architectures and their implications for efficient ML systems.

Representative projects

• Profile algorithm, identifying opportunities for custom XLA/CUDA kernels.
• Implement SOTA architectures (MAMBA, Griffin, Hyena) to research and applied projects.
• Adapt algorithms for large-scale distributed architectures across HPC clusters.

What we offer

• A chance to lead and grow a team of talented engineers in solving some of AI’s most challenging system problems.
• Hands-on experience optimising large-scale distributed ML systems that underpin industry-leading research;
• A front-row seat to the evolution of AI, with opportunities to shape its direction through technical innovation and leadership.