Pengkai Wang's research focuses on two key directions in AI for Science (AI4Science). In AI for medicine (AI4Medicine), he enhances the reasoning capabilities and interpretability of large language models (LLMs) and multimodal LLMs (MLLMs) to support clinical decision-making, aiming to assist physicians in achieving faster and more accurate diagnoses. In AI for partial differential equations (AI4PDE), he develops hybrid approaches that integrate physical priors with deep learning to accelerate simulations of fluid dynamics, thermal transfer, and wind fields, thereby improving efficiency in industrial applications. Both directions aim to bridge advanced AI techniques with domain-specific scientific challenges to create practical solutions.