Marcus Dominguez-Kuhne

About me

I’m a machine learning engineer and researcher working at the intersection of foundation models, decision-making, and real-world autonomy. My work spans reinforcement learning, imitation learning, robotics, and large-scale ML systems. Including papers submitted to IROS and ICLR 2025 .

I earned my B.S. in Computer Science from Caltech with a focus on Machine Learning, and served as a Teaching Assistant for Machine Learning courses. Afterward, I began a Ph.D. in Computer Science at USC (machine learning and robotics) with Prof. Gaurav S. Sukhatme before leaving in 2022 to pursue industry and research opportunities more directly aligned with building applied, real-world AI systems.

In 2025, I researched in the Stanford Multi-Robot Systems Lab , working on risk-aware robotic control using vision-language models (GR00T) focusing on learning safety constraints from human demonstrations and integrating them into control systems using transformer-based models and control barrier functions.

In industry, I was the founding engineer and first employee at SynthAI (YC-funded), where I led engineering and applied research for multi-stage LLM agents — building evaluation pipelines, improving reliability, and designing infrastructure from model training to deployment. Before that, I worked as a Data Scientist and Software Engineer at Amazon Fire TV , developing machine learning models and model-evaluation systems for large-scale recommendation engines.