👋 About Me

I am an AI / ML Researcher & Engineer working at the intersection of representation learning, attention mechanisms, and adaptive language models. My work focuses on understanding how modern models learn and reason, and translating those insights into end-to-end, executable systems.

I currently work as a Research Assistant in Machine Learning and Reinforcement Learning at George Mason University under Professor K.C. Chang, while continuing independent research and systems development in parallel.

I operate as an independent researcher, capable of taking ideas from theory → experiments → systems → communication, while remaining open to collaboration with researchers who share similar rigor and depth.

🧠 Research Direction

My research interests include:

• Self-adapting language models
• Attention and hypernetwork-based architectures
• Representation transfer beyond standard knowledge distillation
• Implicit regularization and generalization in deep models

I am actively developing research contributions aligned with NeurIPS / ICML–style venues, emphasizing originality, clarity, and empirical grounding.

📄 Active Research Work

Latent-to-Parameter Transfer (LPT)
I am developing a research framework that studies how reasoning structure can be transferred from large language models to smaller models via neural-state–conditioned hypernetworks, rather than output-level supervision.

This work explores:

• Conditioning on internal latent states of a frozen teacher model
• Generating low-rank parameter updates (LoRA) dynamically
• Online (per-input) and offline (consolidated) transfer modes
• Improved efficiency in FLOPs while preserving reasoning accuracy
• Near-zero catastrophic forgetting in continual learning settings

The broader goal is to move beyond bandwidth-limited distillation and toward process-level knowledge transfer.

🧑‍⚖️ Academic Service

• Reviewer (3×), Expert Systems with Applications