Thomas Hamori

I am a software engineer at Argonne National Lab working at the intersection of large-scale simulation and applied AI. In the Modeling and Analytics group I architect high-fidelity transportation and logistics models, and I build AI tooling and agentic harnesses that accelerate the researchers who use them. My work centers on verifiable, trustworthy agentic AI: LLM agents rigorously grounded against scientific literature, a simulator, or a formal proof. Separately, I co-developed PyGREET, a Python platform for the GREET life cycle analysis model (Environmental Modelling & Software, 2026), with the Fuels and Materials Analysis group. 

I also serve as the AI Representative for the Strategic Security Sciences division — authoring the Division's first AI training module, advising researchers across all three S³  departments on LLM harnesses, agentic IDE workflows, and multimodal data extraction, and acting as the primary point of contact for over 100 researchers on AI initiatives at Argonne National Laboratory.

My research interests lie at the intersection of agentic AI, applied LLM engineering, and the mathematical analysis, modeling, and simulation of multimodal transportation systems. I'm drawn to one question in particular: when an AI agent tells you something about a complex system, how do you know it's right? I study agentic architectures where language models propose and independent verifiers (simulators, the literature, formal methods) dispose. I'm particularly interested in building secure software that is useful, usable, and used.