- Designed a two-stage malware analysis pipeline using fine-tuned local LLMs to perform malware detection and family-wise classification from sandbox execution logs. - Achieved 97.85% detection accuracy and 97.13% classification accuracy through LLM-based behavioral pattern analysis. - Authored and submitted a research paper titled “MALLM: Malware Analysis with Large Language Models” to SN Computer Science (Springer) — currently under peer review.