Research Kernel Engineer

About the role

We are looking for a Research Kernel Engineer to accelerate the AI Lab’s novel methods by turning them into high‑performance GPU kernels. The role sits at the intersection of cutting‑edge AI research and production‑grade systems, working closely with the AI Cloud platform team.

Key responsibilities

• Design, implement, and optimise CUDA and Triton kernels for new quantisation, sparsity, attention and speculative decoding techniques.
• Perform detailed performance attribution across the serving stack, including roof‑line analysis and identification of bottlenecks.
• Convert published research prototypes that are too slow into production‑ready implementations.
• Build and maintain measurement infrastructure that the research team relies on for benchmarking.

Required profile

• Bachelor’s, Master’s or PhD in Computer Science, Electrical Engineering or a related discipline.
• Hands‑on experience with GPU programming and high‑performance computing in an ML context.
• Proven ability to deliver fast, reliable kernel implementations for real‑world traffic patterns.

Required skills

• CUDA programming
• Triton programming
• GPU kernel optimisation
• Performance attribution and roof‑line analysis
• High‑performance computing concepts
• Machine‑learning systems engineering