Lead a world class team of High-Performance Computing (HPC) Consultants with broad expertise in scientific computing, modeling & simulation. This team supports a broad research community that seeks to understand the behavior of the atmosphere and related Earth and geospace systems by leveraging advanced computer architectures, algorithms, and models.

• Technical Authority & Stewardship of the Aerosicence Disciplines – Aerodynamics, Aerothermodynamics, Rarefied Gas Dynamics, and Parachute/Decelerator Systems – for application within NASA’s Human Spaceflight Programs.• Led the Applied Aerosciences & Computational Fluid Dynamics Branch through multiple programmatic milestones important to NASA’s Human Spaceflight Programs, including development & certification of multiple flight vehicles through direct support of multiple flight tests with in the Orion and Commercial Crew Programs, as well as sustained support for the International Space Station.Notable Milestones Achieved:– Orion Program’s Exploration Flight Test, Ascent Abort Flight Test, Artemis 1 flight readiness.– Commercial Crew Program’s development and qualification activities, from inception throughSpaceX Crew 3 flight readiness.– International Space Station operations & certification of multiple new ‘visiting vehicles‘ andassociated plume-induced aerothermodynamic environment characterization.• Strategic planning & execution of technical work content across a combined government & contractor workforce of ≈35 individuals, with direct responsibility for ≈$10M annual operating budget.• Principal responsibilities for employee recruitment, work assignments, and daily management activities.

• Primary Architect for the CFDLab (UT Austin), Aerolab, and Flight Sciences Labs (NASA/JSC) high-performance computing Linux environments. Led system life-cycle from procurement, through deployment, to operations and retirement for our network, storage, compute, and workstation hardware.– Architected 3 generations of Lustre Filesystems, from 2007 to present. Selected all server components, MDS/OSS width and depth balance, storage subsystem configuration & tuning, deployment, operations, and retirement. Production Lustre experience ranges from v1.6 to v2.14, in a multi-homed InfiniBand/40GbE network environment, including both ldiskfs & ZFS MDT & OST technologies, and active/active (MDT) & active/passive (OST) failover pairing.– Architected/selected 7 generations of Linux clusters, from 2000 to present. Experience dates from very early ‘Beowulf’ 16-node clusters, to (currently) multiple InfiniBand-connected HPE ICE-X/XA systems & federated under a single SLURM interface with a companion Supermicro blade cluster.– Architected a distributed lab environment across the NASA/JSC campus, including network and workstation selection, enabling users’ desktop & software environment to seamlessly match centralized HPC resources, including mounting the same home and Lustre filesystems. This provided a significant efficiency for our use cases, where large files are often required for pre- and post-processing of analyses. Users can develop inputs for PDE analysis directly on the HPC filesystem, and post-process in the same global workspace. Importantly, data transfer for HPC staging was largely eliminated in this setting.

• HPC Systems Administrator with primary responsibility for multiple Lustre, ZFS, NFS, and DMF filesystems, multiple Linux Compute clusters, high-performance end-user workstations, and the software environment for 200+ users.– Responsible for maintenance & operations of Lustre filesystems, including development of custom monitoring tools for use by users and in Ganglia-based diagnostics. For example, aggregate Lustre Jobstats & SLURM job information to quickly identify workflows with un- usual metadata requirements, bandwidth usage, or both. Integrated Robinhood-based usage statistics for rapid user filesystem usage information.– Developed & maintained the common, user-facing HPC software environment on multiple Linux systems. This singly-homed software environment supported 3 clusters and end-user workstations simultaneously, using the Lmod Lua module environment in order to provide transparent support to multiple compiler suites, Python versions, MPI variants (OpenMPI, MPICH, MVAPICH, and HPE/MPT), and requisite development library software stacks simultaneously.– Developed utilities for efficiently dispatching and aggregating Monte-Carlo workflows on HPC systems in a Lustre environment, using an extensible Py4MPI interface, performing local I/O to tmpfs-based filesystems, and aggregating large-scale results to a shared Lustre filesystem. The approach allows for separating small file access from the Lustre filesystem wherever practicable, archiving results instead in a small number of large files.