Naw Safrin Sattar
She/Her
Assistant Professor
Computer Science and Data Science, School of Applied Computational Sciences
Meharry Medical College
Biography
Naw Safrin Sattar, Ph.D., is an assistant professor of high-performance computing in the Computer Science and Data Science Department in the School of Applied Computational Sciences at Meharry Medical College. Dr. Sattarâs research broadly covers high-performance computing, parallel algorithms, scalable large graph mining, big data analytics, machine learning/deep learning at scale, and generative AI. She seeks to support computational scientists and researchers by developing parallel algorithms to solve computationally expensive problems and achieve optimal performance on exascale HPC systems. Dr. Sattar has worked and collaborated at four Department of Energy national labs: Oak Ridge National Laboratory, Pacific Northwest National Laboratory, Lawrence Berkeley National Laboratory, and Los Alamos National Laboratory. These experiences provided her with the opportunity to work on different HPC systems. Collaborating at Berkeley Lab, she developed an optimized distributed parallel algorithm for dynamic graphs, the first MPI-based distributed community detection algorithm for dynamic graphs. During her postdoctoral research at ORNL, she worked on multiple projects, including Graph500 Benchmark on Frontier Supercomputer, Large Scale Graph Analytics for Heterogeneous Computing Platforms, and Machine Learning assisted HPC Operational Data Analytics. She serves as a technical committee member for ACM and IEEE conferences: Supercomputing (SC), IPDPS, Big Data, ISC, and others.
Degrees Earned
Ph.D. in Computer Science/ 2022 M.Sc. in Computer Science/2019 B.Sc. in Computer Science and Engineering/2016
Research Areas
Computer Science; Data Science; Machine Learning/AI; other
Research Interests
Since my undergraduate studies, I have been enthusiastic to work with Big Data. High Performance Computing (HPC) is one such arena closely related to Big Data. Learning the aspects of HPC, I aspire to become a prominent researcher and educator focusing on HPC, AI and Big Data. My research area intersects Big Data Analytics, High Performance Computing, Machine/Deep Learning, Generative AI, Large Language Models (LLMs), Parallel Algorithms, Large Graph Mining, Multi-core Architectures, Distributed Systems, and Security. I have been working on HPC since 2015 to present, with an experience of 9+ years. My research focuses on developing scalable parallel algorithms in various application domains using different HPC techniques. I also work on the scalability of neural networks on GPUs. I develop different prediction ML models and use LLMs for HPC Operational Data Analytics. I work on mining and analysis of both static and dynamic big social and information networks by designing parallel algorithms using HPC techniques. I am the lead author of 15 out of 22 published peer-reviewed full-length conferences / journals [Citation: 337, h-index: 9, i10-index: 9]. My collaborations at 4 national labs: Oak Ridge National Laboratory (ORNL), Pacific Northwest National Laboratory (PNNL), Lawrence Berkeley National Laboratory (Berkeley Lab), Los Alamos National Laboratory (LANL) is a testament of my HPC expertise in handling large-scale data and provided me the opportunity to work on different supercomputers. This experience has provided me with a strong foundation in the computational methods essential for research in engineering and science disciplines. My research agenda centers on interdisciplinary computational science at the nexus of High-Performance Computing (HPC), advanced AI modeling (including Large Language Models, or LLMs), and scalable Large Graph Analysis. Specifically, I focus on developing and implementing novel parallel algorithms to model and interrogate complex systemsâsuch as geospatial data, urban infrastructure, biological networks, or social equity graphsâto extract actionable insights. My goal is to build computationally efficient and equitable AI tools that leverage national lab-scale resources to solve national challenging problems.
Topical Areas
Applied Computer Science; Artificial Intelligence and Intelligent Systems; Atmospheric Sciences; Civil Engineering; Climate and Global Dynamics; Computer Science; Environmental Biology; Environmental Biotechnology; Environmental Engineering; Geology and Solid Earth Sciences; Health Sciences; Other Biological Sciences; Other Computer and Information Sciences; Performance Evaluation and Benchmarking
Research Synergy
I keep my research area open for an interdisciplinary scientific domain where I can incorporate my HPC, Scalable AI, Parallel Algorithms, and Large Graph Modeling expertise. Complex urban systems and infrastructure resilience are key areas I want to explore. My expertise in Large Graph Analysis can be directly applied to modeling city infrastructure (e.g., transportation, utility networks) as complex networks, allowing for HPC-driven simulations of failures, cascading events, and optimization of resource allocation. I am particularly drawn to NAIRR projects focused on developing and training geospatial multimodal foundational models. With my expertise in Scalable AI and Graph Modeling, I can contribute to integrating complex, dynamic Graph-based data representations, leading to a more enriched and efficient foundational model for environmental and Earth science applications. Given that my student team possesses prior Biology/Health domain expertise, I am also enthusiastic about applying AI/HPC, complex network analysis, and parallel algorithms to large-scale biological or health datasets. This aligns with national efforts to address health disparities by developing scalable AI models for precision medicine or public health monitoring. I specifically aim to utilize the SRP program's resources to apply parallel algorithms, complex network analysis to large-scale datasets, potentially tackling problems related to scalable AI models, complex urban systems, environmental, or health domains.
Motivation
My connection to this SRP program is deeply personal. I was a Past SRP Student Fellow, interning with Berkeley Lab in Summer 2019. The SRP program has shaped my career in a meaningful way. That experience was transformational, defining my pathway and fostering a profound interest in national lab research, which I subsequently pursued through a summer internship at Los Alamos National Lab (2021) and a three-year postdoc at Oak Ridge National Lab. Now, as an Assistant Professor of High-Performance Computing at Meharry Medical College, a leading HBCU dedicated to serving the underserved, my primary goal is to close the loop: to give back by providing this same life-changing pathway to the next generation of HBCU scholars. I want to brighten and show the career path to HBCU students through this SRP program. It will provide the participating students with a life-changing experience and provide them with the opportunity to reflect on their long-term career goals. The students in my department are from diverse backgrounds, and shaping their careers by learning advanced technologies through the Master's in Computer Science and Data Science, and Master's in Bioinformatics programs. I have proactively selected a team of highly motivated Master's students in Bioinformatics who have taken my data science foundational programming course. These students, who come from diverse, non-traditional computational backgrounds, have demonstrated exceptional proactiveness, integrity, and enthusiasmâall key indicators for success in high-intensity research. This program will be essential preparation for their capstone research and future careers. It will provide them with a critical opportunity to gain direct exposure to advanced, NSF-funded NAIRR and HPSF projects being conducted at national laboratories. This experience, which addresses computational problems far outside their current curriculum, will enable them to grow into interdisciplinary researchers. HBCU students often face resource disparities. This program directly exposes a student to HPC, Big Data, and advanced AI frameworks, skills typically only accessible at R1 institutions. This program will help my student team to prepare them for competitive PhD programs and increase their interest in national lab research and advanced AI. The SRP program provides a three-way benefit to the project mentors, faculties, and students. As an early-career faculty member specializing in High-Performance Computing, Large Graph Analysis, Parallel Algorithms, Complex Network Modeling, and Scalable and Generative AI, I am eager to expand my research capabilities to solve critical problems of the NAIRR and HPSF project initiatives. I want to expand my knowledge in the cutting-edge NAIRR and HPSF projects. This participation will expand my collaborative network with national lab mentors, which is vital for securing joint funding and strengthening my institutional portfolio for future NSF proposal submissions. Besides, as a faculty member, successfully mentoring a student from Biology into high-level computational science will enhance my capability to collaborate and mentor across different academic departments. From an institutional point of view, the impact is profound. Participating in this program will provide the HBCU masterâs students with basic coding skills to successfully participate in scientific computing projects utilizing HPC platforms, work with Big Data, and AI. Successful completion of the SRP project will also produce quality results for a high-impact publication. The student elevation is a highly competitive advantage for my institution. By successfully mentoring a student team through a rigorous AI/Big Data/HPC project in the SRP program, we can create a documented, reproducible training pipeline that integrates the best practices of the national laboratories that we can learn from the project PIs. This pipeline can be immediately deployed to mentor future masterâs students at Meharry, effectively building sustainable institutional capacity in advanced computational science. This student elevation is ensuring our HBCU produces Master's and PhD graduates capable of contributing to and leading national research initiatives. Along with my professional career development, I want to participate in building this institutional capacity for Meharry Medical College through the SRP Participation. As a team, we are ready to dedicate our best effort to utilize this summer experience to drive research excellence, mentor the next generation of researchers, and ensure Meharry Medical College plays a central role in shaping the HBCU scholarsâ careers with advanced AI and HPC.
Supervising Students Plan
During my postdoctoral research at Oak Ridge National Lab, I had the opportunity to work with 2 summer interns for a 10-week period similar to the SRP program timeline. I mentored a post-bachelorâs and a bachelorâs student through DOE-funded internship programs, and published several posters and a conference paper based on those internship projects. I used to decompose the project into multiple subtasks for each week, and observed their weekly progress. We met twice a week at the beginning and end of the week. These meetings helped keep track of the work of the previous week, ongoing tasks to accomplish in the current week, and how much work had been accomplished by the end of the week. I am responsive to emails whenever they need help. I always advise them to reach out early if they are stuck somewhere in between, and not to wait for regular meeting times. As my previous student interns were from a Computer Science background, my supervision strategy will be slightly different than before. I will continue the same pattern for weekly meetings. My supervision strategy for my student team with two master's students with basic coding and strong domain analysis skills is to implement the "Analyst-to-Scientist" Model. This accelerates their transition from simply analyzing data (the biology/R/Python background) to designing and executing large-scale computational experiments (AI/Big Data/HPC). While decomposing the problem for students, my focus will be on system complexity and networks. Whether it's genes or traffic, the core computational challenge is scaling, graph theory, and prediction. The students should see their existing knowledge as a superpower, not a limitation, enabling rapid adaptation to new data domains (geospatial, social, urban, etc). I will discuss and define the simplest end-to-end Big Data/AI/ML workflow achievable in 4-6 weeks. This will include generating subtasks from a task and ensuring an end-to-end workflow within the project scope. This will ensure a concrete result, even if the later weeks are exploratory. This will build confidence and provide a clear deliverable. This will prevent the students from experiencing anxiety associated with open-ended, complex tasks. I will make sure the students focus on the NAIRR resources or High-Performance Computing environment as the primary workspace from Day 1, and no local development will be encouraged. It will eliminate the technical "scaling gap" later and will force immediate proficiency with shell scripting and job management (SLURM) or other tools they need to use in the project. I will encourage them to maintain a code repository from the very beginning, so it is easily shareable if they face any problems. Instead of exhaustive training, I will guide them to custom computational toolkits for specific project needs (e.g., only the specific functions needed in PyTorch Geometric). This will foster rapid, self-directed learning âessential for an independent researcher. Finally, for the last weeks (Week 8-10), the student should be able to interpret the computational results into a compelling narrative and presentation. The deliverable is a fully documented code repository and a final presentation for the best computational approach for the problem they studied.
Student Merit
My selection of my student team is based on their exceptional performance and proactiveness demonstrated during the Fall 2025 Computer Programming Foundations for Data Science (Python and R) course that I am teaching. They are always attentive and responsive to the class lectures and assignments. The students are not from a Computer Science background. They are quick learners and are progressing well in learning Python and R. They come forward and ask questions if they get stuck at some point, which shows their enthusiasm to learn and grasp more. They are responsive to emails. They have shown their integrity in time management and work hard to complete the assignments/projects provided to them. A key factor was their demonstrated interest in my research area of HPC, Big Data, and ML/AI, which led them to approach me after a department presentation. They just started their Masterâs program, and they are proactive about their capstone research project from the very first semester, which they will complete by the end of their Masterâs degree. Their proactiveness and enthusiasm made me identify them as potential students for my student team. They have shown interest in working in my research area, and this internship will prepare them to progress well in their Masterâs research program and publish papers. By the time they complete the summer internship, they will start the second year of the masterâs program and start working on their capstone project in Fall 2026. This summer internship is an ideal preparatory step to ensure they publish papers and progress successfully. Given their expressed interest in pursuing a PhD program after their Masterâs, this intensive computational experience will be foundational to their long-term research career.
Lightning Talk Title
Scaling the Complex: HPC-Driven Graph AI for NextGen Scientific Computing
Keywords
HPC; Multi-GPU; Parallel Algorithms; Scalable AI; Generative AI / LLMs; Large Complex Network Modeling; Urban Systems; Geospatial AI; Health Informatics;