Murali Emani

Murali Emani is a researcher affiliated with Argonne National Laboratory, IL, USA. His work focuses on High-Performance Computing (HPC), machine learning, and artificial intelligence acceleration. Emani holds a PhD in Computer Science from the University of Edinburgh (2015). He specializes in optimizing large-scale systems, including performance evaluation of AI accelerators, transformer models, and neural architecture search. His research bridges HPC infrastructure with AI applications, addressing challenges in resource allocation, inference efficiency, and cross-architecture benchmarking.

Key contributions include frameworks for GPU memory optimization (XUnified) and holistic performance evaluation of large language models across diverse hardware (e.g., BaKlaVa, Centimani). He collaborates extensively with institutions like the University of Chicago, NVIDIA, and MLCommons on initiatives such as DeepSpeed4Science and GenSLMs for genome-scale language models. His work frequently appears in top venues like IPDPS, SC, and Euro-Par.

Emani’s research emphasizes practical system-level innovations, with applications spanning bioinformatics (SARS-CoV-2 evolutionary analysis), protein design, and exascale computing. He advocates for FAIR principles in HPC data management and contributes to open benchmarks like MLPerf HPC.