Xinyu Lian wins Microsoft Fellowship

Xinyu Lian, second-year CS PhD, has been selected as a recipient of the 2026 Microsoft Research Fellowship. Lian is a member of CS assistant professor Minjia Zhang’s Supecomputing Systems and AI Lab (SSAIL), which focuses on building efficient and scalable systems for modern machine learning workloads. Lian submitted his proposal to Microsoft’s Research Team under the research challenge, “Systems, Algorithms, and Rigorous Applications of Reinforcement Learning Post-Training.” The fellowship, open to faculty, PhD and postdoctoral students, creates collaborative opportunities between Microsoft Research and scholars regarding “research challenges that advance scientific understanding, drive innovation and deliver societal benefit.” The highly competitive fellowship received thousands of proposals.

Lian’s proposal focused on building an efficient and scalable system stack for reinforcement learning (RL) post-training of large language models, with a particular emphasis on agentic tasks such as coding. He noted that though RL post-training can substantially improve model capabilities, it remains constrained by major systems challenges, especially in multi-step, tool-using coding workflows. To address these issues, he proposed developing a unified post-training ecosystem that combines system-level innovations with the goal of delivering a production-ready, open-source pipeline that enables scalable RL training and improves the performance of real-world models.

“There is often a gap between academia and industry, particularly in AI, where the field evolves rapidly,” Lian said. “This fellowship will allow me to collaborate with Microsoft’s excellent research scientists to better understand industry-side challenges and develop practical solutions.”

Zhang, whose group recently received a best paper award at SC, noted Lian’s already-impressive resume so early in his academic career.

“These fellowships are usually awarded to more senior PhD students,” Zhang said. “It is quite rare for someone in the early stages of their PhD to receive this recognition, which speaks to both the originality and maturity of his work."

Lian was also awarded the Amazon AI PhD Fellowship in 2025.

Lian’s trajectory seems to point toward impact-driven research, but regardless of his path, Zhang sees Lian’s work continuing to make an impact in both academia and industry. His proposal to Microsoft was certainly proof of that.   

“The solution he developed was adopted by industry and has made a good contribution to the open-source community,” Zhang said.

Lian recognizes that many real-world challenges cannot be solved within academia alone and require close collaboration with industry.

“I want to solve real-world challenges and develop impactful solutions that can be broadly used,” Lian said.

The fellowship runs March 2026 and through June 2027.

Grainger Engineering Affiliations

Minjia Zhang is an Illinois Grainger assistant professor of computer science in the Siebel School of Computing and Data Science.