Nikos is a Research Scientist in the Machine Assisted Cognition team at TRI. He is interested in developing new AI models that can collaborate closely with humans, each building on the strengths and shoring up the weaknesses of the other. This requires greater interpretability of AI models as well as insights from behavioral science. Nikos has done some theory work on bridging the gap between conventional machine learning models and logical reasoning, which may be useful to provide scaffolding for interpretable AI models.
Before joining MAC, Nikos worked on research projects to apply machine learning to various verification and validation topics, such as adversarial stress testing with reinforcement learning, fault detection in perception systems, and automatically synthesizing lemmas to simplify proofs by a logic engine. His PhD in Electrical and Computer Engineering from Carnegie Mellon was on heuristics to design and verify control systems with a logic engine.