LG AI Research — Vision Lab (Intern)
Deep Document Understanding Team: Developed OCR model for ChatEXAONE.
EXAONE-VL Post Training Team: Worked on EXAONE-VL agentic data generation and SFT.
Hello, I'm Seohyun Lee!
Research Interest
Building reliable VLMs for multimodal reasoning and agentic interaction through post-training and inference-time reasoning.
- Agentic VLMs for complex multimodal task solving
- Prompt optimization for multi-agent VLM workflows
- Long-context multimodal understanding and reasoning
- Post-training and inference-time optimization for reliable VLMs
Experience
