I'm a Machine Learning Engineer working on ML systems, particularly inference optimisation and GPU programming. I'm interested in how hardware, algorithms, and systems design come together to make models run faster and more efficiently.
Previously, I did my Master's in Machine Learning at the University of Nottingham, where I mainly worked on computer vision problems involving hyperspectral imagery under Prof. Michael Pound.
Selected Work
ThunderKittens
Wrote a deep dive into Stanford's Hazy Research ThunderKittens DSL, contributed to the DSL, and implemented and benchmarked an attention prefill kernel against FA2/FA3 on H100. My work was later featured in the main repository.
Emulated
Worked on reinforcement learning tasks related to inference @ Emulated.
CUDA GEMM Optimisation
Documented my experiments optimising GEMM kernels on H100, exploring GPU performance and optimisation techniques.
Feel free to reach out!