• Updated the Unified Atmospheric Algorithm (UAA) software within NASA’s Orbiting Carbon ObservatoryScience Software (OCSSW) suite to integrate new data extraction requirements in NetCDF files, facilitating advanced scientific processing.• Conduct performance analysis on scientific computing applications written in Python and C/C++ for the purpose of optimization at scale.

• Led the Modeling, Simulation, and Emulation team, managing day-to-day operations, setting strategicgoals, and providing long-term vision. Proactively proposed new research topics and projects to driveinnovation and team development.• Managed projects by defining scope and requirements, overseeing funding and budgets, and assessing productivity and risk.• Initiated the founding of the ModSim Tool Consortium, a member-driven entity focused on the longevity.and maintenance of open-source modeling and simulation tools.• Conducted High Performance Computing (HPC) scaling and performance research.• Investigated viability of RISC-V for HPC class systems and their applications.• Characterized and evaluated novel and emerging computer architectures including network interconnects, processing in memory (PIM), silicon photonics, and custom accelerator design.• Implemented the first-ever gate-level processor model in SST, a pipelined RISC-V design, using the custom ArchEssentials library.• Developed custom component libraries and models for the Structural Simulation Toolkit (SST).

• Implemented and evaluated various knowledge and property graph analytics.• Performance and scalability evaluations for heterogeneous distributed memory based HPC class systems,including Sierra Systems.• Utilized the YGM asynchronous communication library to evaluate scalefree knowledge graphs on large scale HPC systems.

• Conducted architectural design space exploration for the purpose of performance optimization on novel systems.• Developed and evaluated high performance computing (HPC) applications.• Implemented hybrid applications with multiprocessing and multithreading, utilizing both distributed and shared memory with MPI and OpenMP.• Developed hardware-specific AI/ML algorithms employing communication avoidance and overhead mitigation techniques.• Emphasis on real time streaming and machine learning, on novel migratory thread architectures and execution patterns.

• Conducted exploratory research into the scalability of irregular memory access (sparse data) applications.• Developed hybrid (shared+distributed memory) parallel software of sparse linear algebra.• Conducted performance analysis of HPC-class heterogeneous systems, incorporating GPUs and customaccelerators such as the Intel Xeon Phi Knights Landing.• Investigated experimental hardware implementations of SpMV and SpMM.• Implemented hybrid applications with multiprocessing and multithreading, utilizing both distributed andshared memory with MPI and OpenMP.

• Planned lectures and developed all course materials.• Lectured on topics including multi-processing and multi-threading.• Conducted in-depth discussions on performance analysis and optimization of conventional architecturesincluding the use of GPUs via CUDA.

• Developed distributed and heterogeneous exascale graph analytics.• Utilized the YGM asynchronous communication library to enable and optimize evaluation of astronomy data (big-data) for the purpose of identifying unique objects and identify their orbital paths. Terabytes of data across millions of individual files, requiring distributed computing via MPI. • Focus on higher order network (HONs) and temporal bi-partite matching using Sierra Supercomputingsystems.

• Investigated existing circuit classification techniques as well as developed multi-threaded clip featureclassification tools to aid in similarity detection with reduced runtimes.

• Developed pedagogical materials utilizing GPUs and NVIDIA CUDA.• Conducted research to evaluate the efficacy of teaching concepts such as memory allocation and management, hardware architecture trade offs, and algorithm advantages/disadvantages.

• Developed heterogeneous implementation of hydrogen plasma fusion• Utilized MOSIX cluster management, with OpenCL for CPU and GPU interoperability