Viktor Veselý

Viktor is an AI researcher and machine learning engineer specializing in deep reinforcement learning, ensemble learning, and theory-based machine learning. His research examines double descent and generalization in reinforcement learning, including how model size, training duration, and experience diversity affect the generalization of learning systems. He also studies how wisdom-of-crowds principles can improve learning systems, with a focus on delegation, problem allocation, and the emergence of specialization across groups of models. More recently, his work has begun to focus on continual learning in reinforcement learning, particularly how models can acquire new tasks without forgetting previously learned ones. Beyond research, Viktor has developed production AI systems for automated assessment, demand planning, customer lifetime value prediction, and other data-driven applications.