Kenneth Marino

Dr. Kenneth Marino is a Research Scientist at DeepMind, set to join the University of Utah as an Assistant Professor at the Kahlert School of Computing in Fall 2025. He earned his PhD in Machine Learning from Carnegie Mellon University (funded by NDSEG and NSF GRFP fellowships) and completed his undergraduate studies in Computer Engineering with a minor in Computer Science at Georgia Tech.

Dr. Marino's research focuses on the intersection of Computer Vision, Natural Language Processing, and Reinforcement Learning, with emphasis on:

• Multimodal agents operating on the web, in simulation, and in real-world environments
• Evaluating AI systems and creating high-impact datasets (e.g., A-OKVQA, OK-VQA)
• Incorporating semantic knowledge into end-to-end learning frameworks
• Embodied reasoning through language-guided planners and neural reporters
• Language model distillation for agent supervision
• Human-AI collaboration to improve RL generalization

His work has been published in top venues including ICML, NeurIPS, ECCV, CVPR, and ICLR, with recent trends emphasizing:

• Relational reasoning in LLMs
• Object-centric world modeling
• Continual learning for embodied agents
• Reporter neural networks for agent control
• Legal AI benchmarking (e.g., BriefMe)
• Knowledge graph integration for visual classification

Scientific recognition includes:

• NDSEG Fellowship
• NSF GRFP Fellowship

He has served on the MLD PhD Admissions Committee and as an Area Chair for ECCV 2024, with prior teaching experience at Columbia University (COMS 6998) and Carnegie Mellon University (16-824, 10-401).