William Arndt
Computer System Engineer at Berkeley Lab
Berkeley, California, United States
Summary
👤
Senior🎓
Top SchoolWilliam Arndt is a Computer System Engineer with 13 years of experience specializing in high-performance computing for genomics and bioinformatics. Based in Berkeley and working at Berkeley Lab, he consults with Joint Genome Institute researchers to optimize workflows on NERSC systems and develops bioinformatics software tuned for large-scale HPC architectures. His PhD-trained background in computer science informs hands-on performance engineering—he achieved a 16x speedup for HMMER on Cori and saved an estimated 3.5 million CPU hours annually. William also contributes to scientific computing projects like the MPAS model, where he improved parallel performance by refining halo exchange and threading behavior. He combines deep low-level optimization skills (vectorization, cache/thread control) with practical tooling and support, making complex HPC systems more accessible to scientists. An understated strength is his ability to translate research-scale performance gains into production practices that materially reduce compute cost and time.
13 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy (Ph.D.), Computer Science, Doctor of Philosophy (Ph.D.), Computer Science at University of South Carolina-Columbia
Github Skills (6)
fortran10
parallel-computing10
performance-optimization10
scientific-computing10
computational-fluid-dynamics9
openmp8
Programming languages (8)
Github contributions (5)
MPAS-Dev/MPAS-Model
Nov 2017 - May 2018
Repository for MPAS models and shared framework releases.
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:7 commits, 1 PR in 5 months
Contributions summary:William focused on optimizing the MPAS model's performance by reducing thread barriers and improving data packing efficiency in halo exchange operations. They refactored code related to data exchange, specifically within `mpas_dmpar.F`, to streamline buffer handling. The user also incorporated halo reuse calls within the barotropic subcycle, improving overall computational efficiency. Their contributions involve performance improvements and parallelization within the MPAS model.
pythonmpas
Larofeticus/hpc_hmmsearch
Jan 2018 - Oct 2022
Top level driver addition to a HMMER3.1b2 installation which more efficiently runs many sequence to many model searches on many cores with threading.
Contributions:5 commits, 1 PR, 5 pushes in 4 years 9 months
searchessequencedriverinstallationruns
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.