Raul Torres

Senior Ai Software Engineer @Huawei Technologies Research & Development (UK) Ltd

Cambridge, GB

Analyze the performance of LLM inference engines (e.g. vLLM) running on computing nodes with multiple accelerators, like GPUs or Ascend NPUs (Neural Processing Units)- Design and implement parameter auto-tuners for Triton AI kernels- Investigate the integration of those auto-tuners as MLIR passes of the Triton compilation stack- Contribute in the enhancement of NPU support in open-source projects like Pytorch NPU, Torchair, SGLang, llama.cpp, xLLM, Bench360 and llmfit- Mentor junior engineers into developing robust practices necessary for profiling and benchmarking HPC and AI applications.

Universidad Nacional de Colombia

Master of Science - MS, Computer Science and Systems Engineering

University of Hamburg

Dr. rer. nat. (PhD), Informatik (Computer Science)

Universidad de Nariño

Bachelor's degree, Systems Engineering