- MAS Affiliated Graduate Students
Stephen P. Kaputsos is an AI / ML architect, engineer, and researcher specializing in agentic AI systems, multi-agent orchestration, LLM-based reasoning, and production AI deployment. Two Master of Science graduate degrees: MIT (Massachusetts Institute of Technology) with a focus in artificial intelligence (5.0/5.0 GPA); Johns Hopkins University (JHU) in Systems Engineering (4.0/4.0 GPA).
His graduate research at the MIT Media Lab centered on agentic AI systems and autonomous decision-making. He led engineering and research teams in designing and building multi-agent deep learning systems, developed custom multimodal SDKs and frameworks exposing agentic AI capabilities to downstream applications, and built evaluation infrastructure for real-time autonomous agent deployment. He also designed and delivered an AI executive education program for senior U.S. Air Force leadership, deploying the system via AWS-backed web infrastructure for real-time AI delivery at scale.
Beyond MIT, he has architected and owned multiple end-to-end agentic AI systems in production at Verizon spanning LLM-based orchestration, multi-agent coordination, and autonomous decision-making across enterprise and consumer environments. He drove technical direction across cross-functional teams, mentored engineers, and ensured AI systems aligned with organizational priorities and end-user needs, translating complex technical capabilities into measurable business outcomes. He co-invented a published U.S. agentic AI patent (2025/0307542 A1) for an autonomous, context-conditioned model selection architecture.
Stephen has led engineering and research teams through the full research-to-production lifecycle, is a named inventor on a published U.S. agentic AI patent filed by Verizon, and is a published IEEE conference author. He is a graduate of the American Management Association's Strategic Leadership Program and a keynote speaker at MIT RealityHack.