AI Safety for Systems That Adapt

About Me

I am a PhD Candidate in Computing and Software at McMaster University working on AI safety and safe continual reinforcement learning.

My research asks how learning-enabled systems can remain safe when the environment, operating context, and the concrete requirements for safe behavior change over time. I develop adaptive safety mechanisms that connect stable high-level safety objectives to context-sensitive constraints, reason about whether adaptation can happen quickly enough, anticipate emerging risks, and provide runtime protection when learned behavior alone cannot assure safety.

My broader goal is to contribute to a methodology for adaptive AI safety: systems that can recognize when their assumptions are changing, anticipate safety-relevant consequences, adapt when justified, and behave conservatively when reliable adaptation cannot be assured.

Research Trajectory

My current AI-safety work grows from a broader background in functional safety and safety-critical software-intensive systems. I am particularly interested in connecting established safety-engineering ideas with the distinctive challenges created by learning and adaptation after deployment.

Education