Anthropic Research Fellow · AI Security · Agentic RL · Learning Theory
I'm a final-year Ph.D. student in the H. Milton Stewart School of Industrial and Systems Engineering (ISyE) at Georgia Tech, advised by Prof. Tuo Zhao, and currently a Research Fellow in the Anthropic Fellows Program.
My research aims to make large-scale AI systems reliable and interpretable: quantitative threat models for AI safety and security; large-scale RL post-training for LLM agents, from environment infrastructure to training that stays efficient and stable over long horizons; and statistical and optimization theory of diffusion models and deep networks.
Measuring how well safeguards hold up against realistic attacks, such as weight exfiltration under egress limits.
Current project ↓Environments, sandboxes, and rollout infra that keep LLM serving, agent rollouts, and training decoupled.
strands-env ↗Long-horizon RL that stays stable, and reward and oversight signals that hold up under optimization.
Related papers ↓Statistical and optimization guarantees for LLMs and diffusion models.
Related papers ↓How well do egress limits protect model weights from exfiltration?
Agent RL post-training on a 100B+ MoE base.
Memory-efficient LLM optimization and structured MoE compression.
Full list on Google Scholar