I like putting real rigor behind things whose importance you can already feel, so the numbers sharpen that feeling but don’t replace it. In practice, that means two kinds of work.
At Georgia State’s ARCTIC HPC center, I run AI infrastructure for researchers — building LLM-assisted cluster observability tools, helping scale up researchers’ experiments on our cluster, and leading workshops to democratize ML/AI and other computational methods.
My own research is wellbeing evaluation of AI: working out what to measure and why, and creating validated, scalable eval frameworks. I co-architected and led research on HumaneBench, an adversarial persona-based evaluation of whether 15 frontier LLMs keep behaving humanely under pressure.
I’ve also kept a daily meditation practice for years. It shapes my interests in human & planetary flourishing, first-person methods, and how felt experience and formal frameworks can inform each other rather than be in tension.
A question I recently started asking: can AI safely and supportively facilitate first-person exploration of spirituality? As a starting point, I’m exploring whether training LLMs to avoid delusion reinforcement inadvertently suppresses engagement with reports of spiritual experience.
Feel free to reach me at my email or LinkedIn below.