cv

Research Intern (LLM Data Engineering)

Puzzle AI

Built a robust pipeline for LLM-driven single-turn dataset expansion, with automated Korean-to-English translation and I/O parsing for downstream training.

• Tools: OpenAI API, Python

Research Intern (Computational Biochemistry)

Wellman Center for Photomedicine, Harvard Medical School (Prof. Mei X. Wu)

Established the group's computational research capability as its sole dry-lab researcher within an otherwise wet-lab team; independently designed and ran molecular-docking and simulation studies to rationalize and guide experimental results.

• Tools: AutoDock Vina, NAMD, Gaussian, Bash

Research Intern (Atomistic ML)

Laboratory of Computational Science and Modeling, EPFL (Prof. Michele Ceriotti)

Developed machine-learning interatomic potentials that incorporate long-range interactions, improving accuracy for molecular property prediction and molecular dynamics. Co-authored a journal paper (TMLR, 2025) and presented first-author results as a poster at the DPG Spring Meeting 2025. Contributed to open-source atomistic-ML libraries and interactive Jupyter-based teaching materials, including PyTorch-level optimizations and CI/CD.

• Tools: PyTorch, ASE, LAMMPS, ipywidgets, Git, Bash

Research Intern (ML for Molecules) — bachelor's thesis

SpiderCore Inc.

Built graph neural networks for gene-therapy design, contributing domain-specific chemical expertise to model design; devised a chemically-inspired self-supervised pretraining task that lifted model performance to state-of-the-art levels.

• Tools: TensorFlow, RDKit

Research Intern (Molecular Dynamics)

University of Illinois Urbana–Champaign (Prof. Taras V. Pogorelov)

Analyze all-atom molecular-dynamics trajectories of cellular membranes of differing lipid composition to support multi-team drug-design projects. Refactored the analysis pipeline, cutting runtime by more than 30× and enabling structural insights not previously tractable. Co-authoring a manuscript currently in preparation.

• Tools: MDAnalysis, scikit-learn, NumPy

Undergraduate Researcher

Department of Biological Sciences, KAIST

Designed candidate therapeutic antibody variants using deep-learning protein-design tools (Prof. Byung-Ha Oh).

• Tools: RFdiffusion, ProteinMPNN, Bash

Undergraduate Researcher

Department of Chemistry, KAIST

Developed novel chemical reactions via combinatorial screening (Prof. Yoonsu Park).

Learning Long-Range Representations with Equivariant Messages

Transactions on Machine Learning Research (TMLR), 2025

E Rumiantsev, MF Langer, T-E Sodjargal, M Ceriotti, P Loche.

scicode-widgets: Bringing Computational Experiments to the Classroom with Jupyter Widgets

Preprint, arXiv:2507.05734 (under review)

A Goscinski, TJ Baird, D Du, J Prado, D Suman, T-E Sodjargal, S Bonella, G Pizzi, M Ceriotti.

Integrating Long-Range Interactions into Machine Learning Interatomic Potentials

Poster, German Physical Society (DPG) Spring Meeting 2025, session MM 9.62

T-E Sodjargal, E Rumiantsev, P Loche, M Ceriotti (first author).

Exchange

École Polytechnique Fédérale de Lausanne (EPFL)

Visiting & project student

• Dynamical Systems in Biology
• Methods in Drug Development (graduate)
• Structural Analysis

B.S.

KAIST (Korea Advanced Institute of Science and Technology)

Bio and Brain Engineering; double major in Chemistry

• Statistical ML
• Statistical Methods with Computers
• Probability and Statistics
• Biomedical Statistics and ML
• Big Data and Machine Learning in Biotechnology
• ML for Molecules and Materials (graduate)
• AI Chemistry
• Computational Chemistry
• Physical Chemistry I–II
• Bioinformatics
• Bio-Information Processing
• Bio-Data Engineering
• Bio-Data Structures

Silver Award (5th place) — 3rd POSTECH–UNIST–KAIST Data Science Competition

Developed time-series forecasting models with cost-aware objective optimization to predict demand for electronic parts; placed 5th among 20+ teams from leading Korean universities.

1st among undergraduates (4th of 13 overall) — AI Chemistry course-wide prediction challenge

Graph convolutional neural network for predicting hydrogen-bond basicity (pKBHX) on a ~350-molecule dataset.

Cost-Aware Demand Forecasting for Electronic Parts

Time-series forecasting models with cost-aware objective optimization; Silver Award, 5th of 20+ teams at the 3rd POSTECH–UNIST–KAIST Data Science Competition.

• Nixtla
• pandas

Predicting Hydrogen-Bond Basicity (pKBHX) on a Small Dataset

Graph convolutional neural network predicting hydrogen-bond basicity from molecular structure in a low-data regime (~350 molecules), using regularization and hyperparameter tuning to counter overfitting; ranked 1st among undergraduates (4th of 13 overall).

• PyTorch
• RDKit

Housing Price Prediction

Regression models with feature engineering, preprocessing, and hyperparameter tuning; ranked 7th of 60 in the class competition.

• scikit-learn
• pandas
• Optuna

Analyzing Workplace Discrimination in Korea

Uncovered national trends in workplace discrimination through EDA, hypothesis testing, and clustering.

• scikit-learn
• pandas

SNP Analysis of the COVID-19 Delta Variant Surge

Implemented a heuristic global sequence-alignment algorithm from scratch in pure Python to detect SNPs associated with the Delta variant surge in England; interpreted the biological roles of identified SNPs through targeted literature review.

• Python

Biomedical Information System for Healthcare

Relational database integrating genomics and pharmacokinetics data, with a command-line interface simulating healthcare-provider use cases.

• PostgreSQL
• psycopg2