Last updated: October 6, 2025

Mingi KANG

Computational Chemistry & Machine Learning

kangmg@korea.ac.kr
github.com/kangmg
kangmg.github.io
Research Interest

My research interests focus on understanding chemical phenomena at the microscopic level through computational chemistry methodologies and accelerating these investigations using AI. Based on this fundamental understanding, I aim to explore and design novel materials and catalysts.

Ab initio
Molecular Dynamics
ML Potential
Machine Learning
Screening
Catalyst
Materials Design
Technical Skills

Computational Chemistry

Ab initio calculaton: Psi4, PySCF, GAMESS-US, Gaussian, ORCA, CP2K etc.
Molecular Dynamics: OpenMM, ASE
Modeling & Simulation: scikit-learn, pytorch, Training ML potentials

Scientific Skills

HPC server & scheduling e.g. PBS, SGE
Dataset handling (HDF5, SQL)
Data visualization

Development Skills

Workflow automation
Python packaging & deployment
Git, Cloud service, API
Research Experience

Bachelor's Thesis

March 2025 - Present

Lab of Ultrafast Spectroscopy, Korea University (Sejong) • Prof. Jae Yoon Shin

  • Investigated dynamic behavior of ionic liquids in porous filters (Polyethersulfone and Anodisc)
  • Performed molecular dynamics calculations using force fields and fine-tuned ML potentials
  • Built simulation systems and analyzed dynamic trajectories to determine diffusion coefficients

CURT Research Program

July 2025 - Present

Laboratory of Inorganic Chemistry, Korea University (Sejong) • Prof. Ho-Jin Son

  • Studied reaction mechanisms of homogeneous transition metal catalysts using DFT and MLP methods
  • Computed Gibbs energy profiles through various levels of DFT
  • Developed automated job submission scripts and established workflows for Gibbs energy profile calculation

Winter Internship

January 2025 - February 2025

The Meta Lab, KENTECH • Prof. Geun Ho Gu

  • Fine-tuned universal MLP models for heterogeneous catalyst systems
  • Reproduced surface adsorption energies using MLP and compared results with DFT

Project Semester

July 2024 - December 2024

Lab of Ultrafast Spectroscopy, Korea University (Sejong) • Prof. Jae Yoon Shin

  • Collected and preprocessed organic molecule dataset (10K samples) using the PubChem API
  • Performed large-scale ab initio calculations (50K) and built normal mode perturbed molecular dataset
  • Fine-tuned ANI Machine Learning Potentials (MLP) to improve accuracy in transition state regions
  • Conducted MLP-based molecular dynamics simulations of carbon polymerizations and acetylene annulation reactions
  • Developed Python package for MLP-based distortion interaction analysis calculations
Education

B.S. in Advanced Material Chemistry

March 2019 - Present

Korea University (Sejong)

GPA: 4.12 / 4.5

Projects

Distortion Interaction Analysis

  • Implemented ASE interfaces for distortion interaction analysis calculations
  • Enabled flexible computations and streamlined computational workflows

Molecule Aligner

  • Developed a Python package for merging, interpolating, and aligning molecular trajectories
  • Enabled construction of smooth, multi-step reaction pathways from mixed molecular inputs

Molecular Visualizer

  • Developing molecular visualization tools including ASE-native viewer aseview and OverlayMol
  • Supports visualization of molecular normal modes, overlay diagrams, and animations with publication-ready quality

Image to Music Recommendation Service

June 2024
  • Proposed the project topic and managed its scheduling, planning, and overall design
  • Collected various types of data using APIs and web crawling, and performed data preprocessing
  • Designed the service architecture, developed, and deployed the website using Streamlit

Deepfake Voice Detection

Sep 2024 - May 2025
  • Proposed a strategy for generating deepfake voice data
  • Preprocessed audio data by adjusting sampling rates and converting file formats to wav
  • Transformed audio data into Mel spectrogram for modeling

ASE Community Code Development

  • Implemented ASE calculator interfaces for g-xTB, PySCF, MLatom, and XequiNet
  • Developed optimizer wrappers for asemcd and geomeTRIC to enhance integration of ASE-centered workflows

randatoms

  • Developed a molecular filtering package using pickle-based metadata caching
  • Used indexed HDF5 storage to enable fast random access and optimize I/O performance for large-scale molecular datasets
Scholarships

Chi-Woo Lee Scholarship

Korea University • 2025

KRW 5,000,000

Academic Excellence Scholarship

Korea University • 2025

KRW 2,350,000
Awards & Honors

Excellent Research Report Award

S-CURT Research Program, Korea University • 2025

Research Grants

Project Semester Research Grant

Korea University • 2024

KRW 1,500,000
Additional Activity

Data Science Bootcamp

February 2024 - July 2024

Completed a 6-month intensive course on data science and machine learning

Military Service

Sergeant

April 2021 - October 2022

Country's Army