Cmptl and Data Sci Rsch Spec 3 - 141146

DEPARTMENT OVERVIEW:

The Mission of the San Diego Supercomputer Center is to translate innovation into practice. SDSC adopts and partners on innovations in industry and academia in the areas of software, hardware, computational and data sciences, and related areas, and translates them into cyberinfrastructure that solves practical problems across any and all scientific domains and societal endeavors. Cyberinfrastructure refers to an accessible, integrated network of high-performance computing, data, and networking resources and expertise, focused on accelerating scientific inquiry and discovery. With more than 250 employees and $30-50M of revenue a year, SDSC is a global leader in the design, development, and operations of cyberinfrastructure.

SDSC supports hundreds of multidisciplinary programs spanning a wide variety of domains, from earth sciences and biology to astrophysics, bioinformatics, and health IT. SDSC presently operates multiple large HPC systems ranging from a 120k x86 CPU core general purpose system to a system explicitly designed for Artificial Intelligence and Machine Learning, and a nationally distributed system open for all of academia to integrate with. SDSC offers research data services across the entire vertical stack from universally scalable storage to consulting services on FAIR, Big Data, and AI. SDSC offers a rich set of cloud services both on-premise, in the commercial cloud, and as hybrid services across both.

SDSC has three geographic scopes, a national scope supporting cyberinfrastructure for the entire US research and education community, a California scope with a special focus on convergence research that addresses the three dominant threats to CA: Drought, Fire, Earthquakes, and a campus scope focusing on advancing the global impact of SDSC by advancing the research objectives of the UC San Diego faculty, researchers, and students. SDSC impacts researchers at scales from 1,000's to Millions. SDSC annually trains thousands of researchers in cyberinfrastructure tools and software, and supports thousands of individual researchers via Unix accounts on its large HPC systems. SDSC was a leader developing the Science Gateway concept, and continues to be a global leader in its evolution. SDSC operates multiple major such gateways with user communities ranging from the tens of thousands to the millions. SDSC's educational programs includes online courses that have been attended by more than a million students.

SDSC is committed to democratizing access to cyberinfrastructure across all of its geographic scopes. SDSC strives towards a culture that supports our employees to be their best, achieve their goals, and enjoy their lives, both professionally and personally.

The Data Enabled Scientific Computing (DESC) division within SDSC designs and jointly proposes with other SDSC researchers, supercomputing systems in response to tens of millions of dollars call-for-proposals from the National Science Foundation (NSF), various government organizations and UC entities; it responds to calls for proposals for cyberinfrastructure (CI) related research, solutions and support. DESC manages, operates and troubleshoots issues with advanced, leading edge, complex, multi-petaflop and multi-petabyte data intensive supercomputer systems, file systems (Lustre, Ceph, BeeGFS etc.), interconnects (such as InfiniBand, NVLink, Slingshot, ethernet etc.) and CI projects housed at SDSC. Research leaders within DESC submit high performance computing (HPC), high throughput computing (HTC), Artificial Intelligence (AI), CI, data science, computational science, science gateways and scientific software research proposals and acquire funding from NSF, National Institutes of Health (NIH), Department of Energy (DOE), Department of Defense (DOD) and industry. DESC carries out supercomputing, AI, CI, data science, computational science and scientific software research and development projects. This division provides consulting and user support at the national level, at UCSD and UC-wide to researchers and users from academia at various US universities and institutions as well as collaborates with them and industrial users. DESC provides advanced computational science, AI, CI and scientific software support for the national and UC user communities as a part of projects/machines such as the Expanse machine (a six-year ~$38-million project funded by the NSF and enables tens of thousands of users to use HPC, HTC and GPUs), the Voyager machine (a five-year, ~$12-million project funded by the NSF and enables researchers to experiment with and use AI-focused hardware for scientific applications), the PNRP project ( a five-year , ~$12-million project funded by the NSF and enables distributed computing with resources of GPUs, FPGAs and CPUs), Cosmos machine (a five-year ~$12-million project funded by the NSF and democratizes access to accelerated computing), the Expanse2 machine ( currently a two-year $10-million project funded by NSF and will be extended for a total of five-years with additional ~$12-million to enable tens of thousands of users to use HPC, HTC and GPUs), the CloudBank2.0 project ( a five-year, ~$37-million project funded by the NSF to enable usage of commercial cloud resources by academic researchers) and the Triton Shared Compute Cluster (TSCC - which is a UCSD condo cluster for UCSD and external researchers and provides the NIST 800 171 compliant HPC and GPU computing). Various other funded CI research and development, and domain science (e.g. biochemistry, bioinformatics, cosmology, physics, engineering etc.) and AI/ML projects are directed by DESC researchers. DESC staff and researchers are involved with and funded by the NSF funded Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program that coordinates various user support, allocations, and training related activities and operations at a national scale across all the NSF funded supercomputer centers located at multiple universities. DESC staff and researchers are involved with and funded by the NSF funded National AI Research Resource (NAIRR) Pilot project which is a multi-institutional national scale project that aims to connect U.S. researchers and educators to computational, data, and training resources needed to advance AI research and research that employs AI. DESC researchers are Co-PIs, subaward site PIs on national scale multi-organization centers and institutes funded by federal funding agencies such as NSF and DOE at the scale of tens of millions of dollars. DESC researchers and staff are involved as PIs, Co-PIs and Senior Personnel in various HPC/HTC/AI training, workshop, outreach, workforce development and K-12 student programs and associated NSF funded projects at the level of multi-million dollars. DESC researchers set trends on R&D and CI/AI training directions involving combination of HPC, HTC, accelerators, AI, CI, scientific software, and domain sciences. This division stays current with HPC, HTC, accelerators, AI, CI, computational science and scientific software research and technology trends. DESC staff engage with supercomputer vendors (e.g. Dell, Supermicro, Intel, NVIDIA, AMD, IBM, Hewlett Packard Enterprise, Data Direct Network, Aeon Computing, Arista, Cambridge Computing etc.) to remain current with future technologies utilized in supercomputer designs.

The Computational and Data Science Research Specialist (CDSR) applies skills as a seasoned, experienced IT research professional and uses computational, computer science, data science, and CI software research and development principles, with relevant domain science knowledge where applicable, along with professional programming concepts for medium-sized projects or portions of larger projects. The incumbent develops and optimizes a variety of computational, data science, and CI research tools and components; performs research on current and future HPC, data, and CI technologies, hardware and software projects; and works on algorithm development, optimization, programming, performance analysis and / or benchmarking assignments of moderate scope where the tasks involve knowledge of either domain / computer science research requirements and / or CI design / implementation requirements.

As a computational & data science researcher, provides advanced application and user support for SDSC's cutting-edge HPC and AI resources, including the Prototype National Research Platform (PNRP), Expanse, Voyager, and Cosmos. The position involves porting research workflows to these innovative resources, developing operational procedures and tools for reliable usage, and integrating these resources into the production research infrastructure. The incumbent creates documentation and tutorials to support the broader use of advanced HPC hardware, such as FPGAs and high-end accelerators, and assist user support staff in onboarding new users. Additionally, provides advanced support for complex issues related to writing, developing, debugging, profiling, and running applications on Kubernetes clusters with novel resources like FPGAs and Habana Gaudi processors. This includes conducting in-depth analysis of factors impacting user code performance and performing complex troubleshooting in a Kubernetes environment. The CDSR supports the use of machine learning tools on SDSC resources for a diverse research community and advise on software and algorithm choices for data-intensive problems.

The incumbent also collaborates with the systems group to integrate advanced HPC hardware, including FPGAs and high-end accelerators, into the PNRP and Voyager clusters, modifying software stacks to enable network and scientific applications to utilize these innovative resources. This includes developing, testing, and implementing software stack modifications in the production environment for user access. In addition, develop and maintain comprehensive documentation and tutorials on FPGA and innovative hardware/software resources available at SDSC, keeping usage documentation updated, and presenting tutorials to enable broader use and assist user support staff in onboarding new users. Provides user support to help researchers and educators integrate SDSC and NRP provided inference services into their workflows and classes, including use within Jupyter environments. Responsibilities will also include developing and maintaining benchmark suites to test FPGAs and innovative hardware on SDSC Kubernetes clusters, providing advanced support for testing, implementing, and integrating complex domain science applications into workflows on Kubernetes-based clusters using innovative hardware/software components.

The incumbent works closely with SDSC research staff to aid and participate in proposals utilizing SDSC's unique HPC resources, such as Expanse, Voyager, PNRP, and Cosmos, as well as in proposals for future CI resource procurements. This involves running applications and micro-benchmarks to characterize performance in support of proposals, designing and implementing plans to use application test cases to highlight the unique features of proposed systems, including computational features, virtualization, high-performance networks, and I/O components. Additionally, the CDSR collaborates with SDSC project managers to develop and maintain research collaborations and projects utilizing SDSC HPC resources. This position offers an exciting opportunity to work at the forefront of HPC and AI, contributing to cutting-edge research and technological advancements.

For more information, please visit: https://www.sdsc.edu/

• Bachelor's degree in related area and / or equivalent experience /training. Master's degree preferred.

• Advanced skills and demonstrated experience associated with one or more of the following: HPC hardware and software power and performance analysis and research, design, modification, Implementation and deployment of HPC or data science or CI applications and tools.

• Thorough experience working in a complex computing / data / CI environment encompassing all or some of the following: HPC, data science infrastructure and tools / software, and diverse domain science application base.

• Demonstrated broad experience in one or more of the following: optimizing, benchmarking, HPC performance and power modeling, analyzing hardware, software, and applications for HPC / data / CI.

• Demonstrated experience in using AI/machine learning approaches in domain science applications.

• Proven ability to work in a Kubernetes cluster environment with experience using it for either research or education.

• Job offer is contingent upon satisfactory clearance based on Background Check results.