Irene Simó Muñoz

My research focuses on making large-scale computation more efficient and accesible through novel hardware and algorithms, so that scientists in other fields like climate modelling, drug discovery, and astrophysics can run bigger simulations and train larger models without waiting days for results.

Concretely, I work on Processing-In-Memory (PIM) architectures and low-precision arithmetic for high performance computing. PIM dramatically cuts energy consumption and speeds up data-hungry workloads like sparse matrix operations, which are the backbone of scientific simulations and machine learning inference alike. Combined with low-precision arithmetic, which trades a small, controlled amount of numerical accuracy for large gains in speed and memory, these approaches let researchers tackle problems that were previously out of reach on conventional hardware.

Previously, I was a software engineer at the Barcelona Supercomputing Center (BSC) developing the Autosubmit workflow manager for climate modelling. Before that, I was at Georgia Tech as a graduate researcher. I collaborated with Prof. Rich Vuduc in randomized triangle counting and won 2nd place in the ACM Student Research Competition (Graduate) @ SC’24. I also worked in PDE-constrained optimization under uncertainty under Prof. Peng Chen’s supervision.

My research interests include Processing-In-Memory architectures, low-precision arithmetic, Numerical Linear Algebra and graph problems.

In terms of education I have obtained both my Master’s in Computer Science (2024) and my Bachelor’s in Aerospace Engineering from UPC (2022).

Past experiences include an internship at Hewlett-Packard. I since have gravitated towards more academic-inclined positions, collaborating with the Barcelona Supercomputing Center (BSC) for my bachelor thesis, and interning at the International Centre of Numerical Methods in Engineering (CIMNE) among other collaborations with professors.