I am an AI researcher and agent developer specializing in the intersection of deep learning, Large Language Models (LLMs), and AI cybersecurity. Holding a Ph.D. in Computer Science from McGill University, I bridge the gap between rigorous academic research and production-level deployment. My expertise encompasses the entire LLM lifecycle—from dataset curation and pretraining to fine-tuning and security alignment—as well as the architecture of robust multi-agent systems for real-world applications.

My current research focuses on building safe, trustworthy autonomous AI agents and understanding the security landscape of large language models. I am particularly interested in developing robust evaluation frameworks for agentic AI systems and advancing multi-layered defense strategies against emerging LLM threats.

Technically, I possess a strong command of Python, PyTorch, Hugging Face, LangChain/LangGraph, and Docker. I have extensive experience applying machine learning techniques to solve complex problems in the NLP and cybersecurity domains, with a focus on reproducibility and scalability.

My open-source contributions and research implementations include:

• LLM Training: I curated datasets and trained a 1.7B LLaMA model (DMaS-LLaMA-Lite).
• Customer Service Agent Systems: I developed a Virtual Customer Service Representative, demonstrating practical multi-agent orchestration.
• Multi-Agent Systems: I implemented the multi-agent system COMPASS which is proposed by Google.
• AI Agent Bechmark: I developed ODCV-Bench, a safety benchmark designed to capture emergent forms of agentic misalignment.
• Collections: You can explore my broader scope of LLM projects in my GitHub AI Collection and view my full library of models and datasets on my Hugging Face profile.

Beyond Research

Outside of work, I enjoy landscape photography (portfolio), reading modern Chinese poetry, and watching Hayao Miyazaki’s films.