Hi, I’m Hongzhou, an Applied Scientist at Amazon, where I am part of the AGI Foundation team. My current work focuses on the development of large language models (LLMs) with a specialization in mathematical reasoning and problem-solving, both informal and formal (i.e. LEAN). I contributed to the development of Amazon's recently launched Nova models, which are designed to handle advanced multimodal tasks across text, images, and video.

Previously, I worked on diverse projects spanning supply chain management, price modeling, and applications of LLMs across various domains. These experiences have given me a strong foundation in optimization, predictive modeling, and the innovative use of AI to solve real-world challenges.

Before joining Amazon, I conducted research in optimization algorithms, deep learning, and reinforcement learning theory. I completed my postdoctoral work at the Machine Learning Group at MIT under Stefanie Jegelka and earned my PhD from INRIA Grenoble in 2017 under the mentorship of Zaid Harchaoui and Julien Mairal. My doctoral research focused on developing generic acceleration schemes for first-order optimization algorithms, laying a strong theoretical foundation for my subsequent work in applied science.