Summary
Christopher Gagne is a staff research scientist and AI generalist with a decade of experience translating computational neuroscience and cognitive science into production-grade multimodal AI. He has led research and engineering teams at Hume AI to build real-time speech-text LLMs (EVI 1–3) and architected novel RLHF/RLVR pipelines to improve voice realism and dynamic modulation. His academic work under Peter Dayan and at Berkeley blends distributional and risk-sensitive reinforcement learning with Bayesian hierarchical modeling to illuminate decision-making, mental simulation, and psychiatric phenomena. Now at Google DeepMind, he bridges cutting-edge research and productization, combining large-scale data pipelines (>1M hours processed) with principled evaluation frameworks. Less obvious: he has a track record of turning theoretical constructs (CVaR, risk-sensitive Bellman equations) into practical probes of LLM behavior and speech systems. Based in New York, he excels at uniting rigorous theory, scalable engineering, and interdisciplinary teams to push multimodal AI toward human-like conversational capabilities.
10 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy - PhD Cognitive Neuroscience, Doctor of Philosophy - PhD Cognitive Neuroscience at University of California, Berkeley
BS BA Psychology (Neuroscience conc.) Philosophy, BS BA Psychology (Neuroscience conc.) Philosophy at Boston College