We are the Data Science & Artificial Intelligence Lab (DSAIL) at KAIST, led by Prof. Chanyoung Park.

Our goal is to mine meaningful knowledge from complex multimodal data and develop robust, scalable AI solutions to solve real-world challenges across diverse scientific and engineering disciplines. Moving beyond traditional machine learning, two underlying themes of our current research are:

1. Representation & Alignment: How can we extract and align knowledge from diverse modalities (e.g., texts, graphs, molecules, videos) into unified representations, bridging the gap between domain-specific data and Large Foundation Models (e.g., LLMs) to capture deep semantic relationships?

2. Reasoning & Agentic Fusion: How can AI agents logically synthesize extracted knowledge to reason, plan, and execute complex tasks? We focus on customizing these reasoning trajectories to facilitate autonomous problem-solving in underlying target applications.

Our current active research interests emphasize AI for Science, Agentic AI, (Multimodal) Large Language Models, and Data-centric AI, expanding into the following core foundations and applications:

AI Foundations

• Large Language Models & Scientific LLMs
- multimodal LLMs, scientific reasoning, tool use, distillation, video LLMs
• Graph Neural Networks & Relational Learning
- self-supervised graph learning, molecular graphs, relational representation learning
• Multimodal Representation Learning
- graph–text–sequence integration, cross-modal alignment, Relational multimodal data
• Agentic AI & Reasoning Systems
- LLM agents, planning, multi-step reasoning, tool-augmented learning, memory
• Generative Models & Inverse Design
- molecular generation, diffusion models, design optimization
• Robust & Interpretable Machine Learning
- explainability, out-of-distribution generalization, reliable AI

Applications

• AI for Science (Biology, Chemistry, Materials)
- single-cell modeling, virtual cell, perturbation, molecular property prediction, protein–ligand interaction, generative molecule design, spectra-to-structure reasoning, scientific AI agents, materials discovery
• Recommender Systems & User Modeling
- LLM-based recommendation, graph-based recommendation, sequential modeling
• AI for Engineering (Fluid Dynamics & PDEs, EDA)
- neural operators, physics-informed learning, simulation acceleration, design optimization
• Graph-based Real-world Data Intelligence
- social networks, knowledge graphs, relational data mining
• Fraud & Anomaly Detection
- financial fraud, graph anomaly detection, risk modeling
• Time-series & Spatio-temporal Modeling
- dynamic systems, temporal graph learning

Interested?

If you’re interested in joining our lab, send an email with your interests, CV, and transcript to cy.park (at) kaist.ac.kr.