Discovery of Spectrographic Features in Uranium Compounds through Machine Learning CSMD Computer Science and Mathematics Division ORNL
Discovery of Spectrographic Features in Uranium Compounds through Machine Learning
ORNL researchers Phani Marthi and Suman Debnath work on developing and scaling up new EMT simulation software to analyze how power electronics in the electric grid will respond to brief interruptions in power flow. Credit: Genevieve Martin/ORNL, U.S. Dept. of Energy
New software provides advanced grid simulation capabilities

Software Engineering Group

Engineering the next generation of high-quality scientific software.

The Software Engineering Group (SE Group), led by Addi Malviya-Thakur, is dedicated to enabling groundbreaking scientific research by developing and deploying cutting-edge software solutions. The group’s core mission is to advance ORNL's objectives by facilitating the use of High-Performance Computing (HPC) and computational science techniques across a diverse range of scientific domains. This is achieved through deliberate partnering across scientific domains to tackle complex research challenges, conducting research in scientific software engineering, and translating those innovations into practical tools that enhance scientific workflows and accelerate discovery. The ultimate goal is to empower scientific teams with the computational expertise necessary to leverage state-of-the-art computational methods.

The group maintains a comprehensive set of technical capabilities crucial for modern scientific computing. Expertise spans Scientific Software architecture, design and development, full-stack development, machine learning model development and deployment, and the creation of web-based services. A key strength lies in High-Performance Computing and Performance Optimization, including the development of optimized software for HPC systems and performance engineering for efficient resource usage.

The SE Group is pioneering Machine Learning and AI Workflow Engineering, designing and developing ML/AI-enabled scientific workflows. This includes everything from model training pipelines and inference optimization on HPC systems to the integration of AI/LLM agents into experimental steering and code generation. In addition, the group is proficient in architecting complex systems, specifically focusing on Interconnected Science and Federated Systems; building software frameworks that seamlessly integrate various labs, computing, and operational facilities. These technical skills are underpinned by strong disciplines in Security and DevSecOps, automation, workflow optimization, and cloud computing.

The group’s impact is demonstrated through its leadership and contributions to major platform-level software. The SE Group owns foundational tools like the Interconnected Sciences Software Platform, the INTERSECT SDK , and numerous packages for Neutron Sciences. SE group's innovative work has garnered external recognition, including a Best Paper Award at the XLOOP 2024 (SC24) workshop for integrating ORNL’s HPC and Neutron Facilities.

Composed of technical and research professionals, the group brings multidisciplinary expertise spanning scientific software engineering, computational physics, quantum computing, artificial intelligence, machine learning, and applied mathematics to tackle the nation’s most complex and compelling scientific challenges.