Xuefeng Li
I am a Senior Research Engineer at Modelcode.ai, focusing on how AI agents tackle long-horizon tasks, particularly repository-level code modernization. I work across the full agent stack — benchmark and dataset creation, agent harness building, and model fine-tuning — with results published at top conferences.
Previously, I worked at NinjaTech AI, Rokt, and Amazon. I hold a Ph.D. in Computer Science from UNSW Sydney, where I worked on weakly supervised learning and generative models, advised by Alan Blair and Tongliang Liu.