Seunghyun Joo

Seunghyun Joo

Research Associate at Seoul National University | Looking for a PhD position

About Me

👋 Hello, I’m Seunghyun Joo

I am a Research Associate at the Institute of Advanced Aerospace Technology, Seoul National University, where I continue my research with the Aerospace Vehicle Design Laboratory (AVDL). I received my M.S. in Aerospace Engineering from Seoul National University in August 2026. My research lies at the intersection of Fluid Dynamics and Data-Driven Science, with a specific focus on Turbulence Modeling, Physics-Informed Machine Learning, and Adjoint-Based Aerodynamic Design.

🎯 I am currently looking for a PhD position. Please feel free to contact me at chlrh45351@gmail.com.

🔍 Research Interests

My research goal is to develop robust aerodynamic models by integrating physics-based simulations with data-driven techniques.

  • Data Assimilation & AI: Enhancing turbulence models and flow field reconstruction using machine learning and data assimilation techniques.
  • Multi-Fidelity Aerodynamic Analysis: Expertise in Low (VLM, Panel method), Mid (LBM), and High-fidelity (RANS/LES/DNS) CFD simulations.

💼 Experience

  • Research Associate, Institute of Advanced Aerospace Technology, Seoul National University, Sep 2026 – Present

🎓 Education

  • M.S. in Aerospace Engineering, Seoul National University, Sep 2024 – Aug 2026
  • B.S. in Aerospace Engineering & B.B.A. in Business Administration (Double Major), Seoul National University, Aug 2024

📚 Publications

CuAD: Source-to-Source Automatic Differentiation of CUDA Kernels for Discrete Adjoints in Computational Fluid Dynamics

Preprint, arXiv (physics.flu-dyn), 2026

A source-to-source AD tool that writes tangent and adjoint CUDA kernels whose reverse sweep recomputes instead of storing, compared under one compiler with Clad and Enzyme and applied to a complete transonic Euler adjoint. Preprint: arXiv:2610.26694.

Recommended citation: Seunghyun Joo. (2026). “CuAD: Source-to-Source Automatic Differentiation of CUDA Kernels for Discrete Adjoints in Computational Fluid Dynamics.” arXiv preprint arXiv:2610.26694.
Download Paper

A Fully Coupled CFD–Free-Wake Discrete Adjoint for Rotor Design: Retaining Far-Wake Sensitivity

Work in Progress, 2026

The CFD residual and the periodic free-wake equilibrium residual are solved as one coupled system in both the primal and the adjoint, so the far wake is modeled cheaply but its design sensitivity is retained.

Recommended citation: Seunghyun Joo, Kwanjung Yee. “A Fully Coupled CFD–Free-Wake Discrete Adjoint for Rotor Design: Retaining Far-Wake Sensitivity.” Work in progress.

Vortex-Aware Reconstruction Selection for Reducing Vortex Dissipation, Informed by Field Inversion and Symbolic Regression

Under Review, Journal of Computational Physics (JCP), 2026

A vortex-aware (VA) reconstruction selector that reduces the numerical dissipation of under-resolved vortices in compressible flows while keeping the shock-capturing behavior of the baseline scheme.

Recommended citation: Seunghyun Joo, Yoonpyo Hong, Younghyo Kim, Kwanjung Yee. (2026). “Vortex-Aware Reconstruction Selection for Reducing Vortex Dissipation, Informed by Field Inversion and Symbolic Regression.” Under review, Journal of Computational Physics.

Mechanism-Separated Closed-Form Transition Modeling via Field Inversion and Symbolic Regression

Under Review, AIAA Journal, 2026

A closed-form, mechanism-separated laminar-to-turbulent transition model for the Spalart–Allmaras equation, built with field inversion and symbolic regression. Preprint: arXiv:2609.24162.

Recommended citation: Seunghyun Joo, Younghyo Kim, Kwanjung Yee. (2026). “Mechanism-Separated Closed-Form Transition Modeling via Field Inversion and Symbolic Regression.” Under review, AIAA Journal. arXiv:2609.24162.
Download Paper

Data-Driven Algebraic Turbulence Transition Modeling via Hierarchical Separation of Physical Transition Mechanisms (In Korean)

Published in KSAS Spring Conference 2026, 2026

Data-Driven Algebraic Turbulence Transition Modeling via Hierarchical Separation of Physical Transition Mechanisms

Recommended citation: Seunghyun Joo, Younghyo Kim, Kwanjung Yee. (2026). “Data-Driven Algebraic Turbulence Transition Modeling via Hierarchical Separation of Physical Transition Mechanisms.” KSAS Spring Conference 2026, (TBD).

AMADEVS: A Generalized Aircraft Modeling & Analysis Environment for Multidisciplinary Design

Published in AIAA SCITECH 2026 Forum, 2026

A generalized aircraft modeling and analysis environment with role-based components and a dual geometry–energy topology, demonstrated on Mars helicopter sizing, wing design under propeller–wing interaction, and UAM quadcopter handling qualities.

Recommended citation: Jeonghwan Kim, Seongjoong Park, Berenice Chatelain, Seonghyun Jeong, Jaehyun Ryu, Seunghyun Joo, and Kwanjung Yee. (2026). “AMADEVS: A Generalized Aircraft Modeling & Analysis Environment for Multidisciplinary Design.” AIAA SCITECH 2026 Forum, 1687.
Download Paper

Development and Validation of a Low-Fidelity Analysis Tool for Conceptual Design Accounting for Anti-Icing Energy Requirement (In Korean)

Published in KSAS Fall Conference 2025, 2025

Development and Validation of a Low-Fidelity Analysis Tool for Conceptual Design Accounting for Anti-Icing Energy Requirement

Recommended citation: Heesang Shin, Younghyo Kim, Seunghyun Joo, Kwanjung Yee. (2026). “Development and Validation of a Low-Fidelity Analysis Tool for Conceptual Design Accounting for Anti-Icing Energy Requirement.” KSAS Fall Conference 2025, (TBD).

Downwash/Outwash Analysis of eVTOL Configurations Using the Lattice-Boltzmann Method

Published in Asia Pacific International Symposium On Aerospace Technology (APISAT 2025), 2025

Analysis of aerodynamic characteristics of eVTOL configurations using the Lattice Boltzmann Method (LBM).

Recommended citation: Juneyoung Lim, Seunghyun Joo, and Kwanjung Yee. (2025). “Downwash/Outwash Analysis of eVTOL Configurations Using the Lattice-Boltzmann Method.” APISAT 2025.

🛠️ Research Projects & Portfolio

CuAD: Source-to-Source Automatic Differentiation for CUDA

An open-source, Tapenade-style source-to-source AD compiler for CUDA built on Clang LibTooling. It generates exact forward (tangent) and reverse (adjoint) GPU kernels from the original CUDA source, and powers the discrete adjoint of CuDAFLOW.

CuDAFLOW: CUDA Discrete-Adjoint CFD

An FP32 GPU finite-volume RANS/LES solver built around an exact discrete adjoint whose residual derivatives are generated by CuAD — a source-to-source automatic-differentiation compiler for CUDA. Drives gradient-based shape optimization, Cf/Cp inverse design, and Spalart–Allmaras field inversion (FIML) from Python, with the adjoint linear solve in double precision through PETSc. Validated on Schubauer–Klebanoff transitional flat plate, Mach-7 axisymmetric shock/boundary-layer interaction, and the ONERA M6 transonic wing.

✉️ Contact