Define and prioritize technical research bets across large language models (LLMs), multimodal AI, agents, graph machine learning, knowledge graphs, safety, neurosymbolic reasoning, alignment, and evaluation. You will develop methods that represent enterprise data, rules, permissible actions, and business processes as linked, versioned layers, then connect them to learning-based models and agents.