Sourena MOOSAVI

• Migrated and refactored a legacy thermal simulation tool from R to Python to modernize the codebase, applying SOLID principles and design patterns (Factory, Strategy) for scalability and maintainability, and introducing multiprocessing for selected operations which improved performance by 200% on CSV datasets containing up to 100𝑘 rows with multiple columns. (Python, R, Numpy, Pandas, YAML, Multiprocessing). • Enhanced computational tools used for nuclear site simulations by implementing new features and performance improvements in C, Fortran, and R, and integrating external C components into the Python architecture as part of a prototype for future high-performance modules. • Boosted simulation performance through OpenMP parallelization, reducing runtime from 240 seconds to 130 seconds (≈ 45% improvement) following detailed profiling and optimization with Intel-VTune and Intel Advisor. • Designed, deployed, and optimized a PostgreSQL/PostGIS-based geospatial database for large-scale environmental models, achieving a 60% reduction in query execution time through indexing and cutting data load time from 15 minutes to under 2 minutes using batch copy operations. • Authored clear, detailed Software Requirements Specifications (SRS) to align technical solutions with project goals.

• Refactored legacy C++ and Fortran code (CATHARE) for thermal-hydraulic simulations, reducing runtime from 40 minutes to 30 minutes (≈ 25% improvement) by replacing inefficient traversal and update patterns with array-based data structures to improve memory locality and reduce overhead, processing thousands of meshes with tens of thousands of degrees of freedom. • Collaborated with cross-functional teams to enhance codebase scalability, reducing technical debt in a complex scientific system. • Led testing and validation efforts, ensuring simulation accuracy across diverse use cases with zero critical defects. • Developed efficient algorithms to meet strict performance requirements, balancing speed and reliability.