Graduate Student Researcher — AI & Computational Fluid Dynamics
Enabling faster design iterations for nuclear systems
I build hybrid CFD + machine learning workflows in collaboration with amazing team so that we can explore more designs for advanced nuclear systems without compromising safety margins.
I’m a student researcher working at the intersection of CFD, physics, and machine learning, building hybrid models that can predict full flow fields while staying faithful to the underlying PDEs and numerical methods.
- Affiliation: POSTECH
- Course: Ph.D. student
- Supervisor: Prof. Joongoo Jeon
- Laboratory: NINE Lab
I baffle myself by the vastness of cosmos, minuteness of atoms and everything in between.
Research highlight
Fluid intuition, fully resolved
Natural circulation
Flow predictions, 3D, in motion
Surrogate flow fields generated with XRePIT for a 3D natural-circulation case — high fidelity at a fraction of the cost.
Framework
How XRePIT works
Physics-informed ML on top of CFD baselines — predictions that respect the underlying PDEs.
Selected work
Publications
- Shilaj Baral, Youngkyu Lee, Sangam Khanal, and Joongoo Jeon. "XRePIT: A deep learning–computational fluid dynamics hybrid framework implemented in OpenFOAM for fast, robust, and scalable unsteady simulations." Computers & Fluids (2026): 107075. View
- Sangam Khanal, Shilaj Baral, and Joongoo Jeon. "Comparison of CNN-based deep learning architectures for unsteady CFD acceleration on small datasets." Nuclear Engineering and Technology (2025): 103703. View
- Joongoo Jeon, Jean Rabault, Joel Vasanth, Francisco Alcántara-Ávila, Shilaj Baral, and Ricardo Vinuesa. "Inductive biased-deep reinforcement learning methods for flow control: Group-invariant and positional-encoding networks improve learning reproducibility and quality." Physics of Fluids 37, no. 7 (2025). View
Working with
Collaborators
From the notebook