Michael Carilli

Michael Carilli has a wealth of experience in applied cryptography, deep learning frameworks, computational science, and high-performance computing.

From 2010 to 2015, they were a PhD Student at UC Santa Barbara, where they adapted numerical techniques from barrier-crossing theory to polymer self-consistent field theory and other field-theoretic models. Michael also devised parallelization and memory-saving schemes to optimize these techniques for various high-performance computing architectures using CUDA and MPI.

From 2015 to 2017, they worked as a Computational Scientist at ERC, where they ported performance-critical physics routines for rocket CFD code from Fortran to C++ using Kokkos. Michael optimized their code for Nvidia GPUs, Intel Xeon Phis, or multicore vector CPUs, achieving up to 32X speedup over serial Fortran. Michael also performed "code surgery" to interface Kokkos-accelerated C++ routines with large Fortran codebase, emphasizing minimal alteration to existing code and minimal impact on other developers.

From 2017 to 2022, they worked at NVIDIA as a Senior Developer Technology Engineer (Deep Learning Frameworks). Michael focused on making GPU training fast, numerically stable, and easier for internal teams, external customers, and Pytorch community users. Michael contributed Python frontends, optimized CUDA kernels, and C++ backend fixes to improve mixed-precision, multi-GPU, and eager mode user experience.

Currently, they are a Senior Applied Cryptography Engineer at Matter Labs, where they are helping make zkSync the world's first (as far as they know) GPU-accelerated zero-knowledge rollup. Michael is also helping Ethereum scale to thousands of transactions per second without sacrificing decentralization or security.

Michael Carilli has a Doctor of Philosophy (PhD) in Physics from UC Santa Barbara, obtained in 2015. Michael also has a BS in Physics from the University of Notre Dame, obtained in 2010. Additionally, they have obtained four certifications from LinkedIn and Colfax International in 2021 and 2020, respectively.