Ruicheng Ao 敖睿成

Ruicheng Ao Email: aorc [at] mit (dot) edu Google Scholar CV I'm too fruitless. When you see this message I'm likely to being moyuing.

About Me

My last name Ao (敖) = AI × OR — a coincidence that became a career.

I'm a third year Ph.D. candidate in the Institute for Data, Systems, and Society (IDSS) at MIT, working with Prof. David Simchi-Levi and Prof. Thomas Magnanti. My research builds efficient, reliable, and aligned AI systems grounded in optimization theory and statistical foundations. My work explores the multidirectional synergies between Operations Research, Statistics, and LLMs — using OR and Statistics to make LLMs faster and smarter, while leveraging LLMs to solve hard problems in OR and Statistics. I'm open to collaborate with anyone interested in related topics. Before MIT, I was advised by Prof. Zaiwen Wen, Prof. Yuejie Chi, and Prof. Jing Dong. I received my B.S. in Mathematics from Peking University in July 2023.

I'm fortunate to continue working closely with Prof. Jing Dong on theoretical research. I'm deeply grateful for her generous mentorship and support over the years.

Research Interests

My research explores the multidirectional synergies between Operations Research, Statistics, and Large Language Models — using OR and Statistics to make LLMs faster and smarter, while leveraging LLMs to solve hard problems in OR and Statistics.

OR & Statistics for LLMs

• LLM Serving & Inference Optimization: Fluid-dynamics-based scheduling for LLM serving; optimal routing across heterogeneous models; memory-aware batching under stochastic arrivals.

• LLM Evaluation & Human-AI Decision-Making: Evaluation with LLM judges under limited human supervision; bandit algorithms for scalable preference learning; prediction-powered inference combining LLM outputs with gold-standard labels; interactive systems for human decision-making with ML predictions.

• Optimization for Multi-Agent Systems: Resource allocation, communication protocols, and coordination mechanisms for LLM-based multi-agent workflows.

LLMs for OR & Statistics

• LLM Agents for Optimization: Closed-loop LLM agents for diagnosing and repairing optimization models; MDP-based frameworks for evaluating self-correction and tool use; interactive benchmarks for LLM–solver integration.

• Revenue Management & Service Operations: LLM-driven approaches to data-driven pricing, demand learning, inventory control, and service system design.

• Statistical Learning & Experimental Design: Adaptive experimental design; Bayesian online testing for multi-armed decisions; interactive systems for human decision-making with ML predictions.

Summer 2025, I'm working as research intern in Wotao Yin's group and supervised by Xinshang Wang (who sits 16 hours besides me at company each day) at Alibaba, where I won awards twice in math competition. Spring 2025, I worked with Prof. Thomas Magnanti on an industry project concerning EV charger scheduling and installation.

I also continue working with Xinshang Wang on various LLM-related projects, who has been a constant source of discouragement — which, oddly enough, keeps me going.

Awards & Honors

• Finalist, POMS College of Service Operations Management Best Student Paper Competition, 2026

• Finalist, INFORMS Applied Probability Society Best Student Paper Prize, 2025

• Finalist, INFORMS Revenue Management and Pricing Section Jeff McGill Student Paper Prize, 2024

• Excellence Award, Alibaba Global Mathematics Competition, 2024

• Excellence Award, Alibaba Global Mathematics Competition, 2022

• Gold Medal (1st Place), S.-T. Yau College Student Mathematics Contest, 2022

• Silver Medal (2nd Place), S.-T. Yau College Student Mathematics Contest, 2021