Hello! I'm
Currently an intern with the Hunyuan AI Data Department at Tencent, working on frontier benchmark research.
I received my MSc in Financial Technology from the School of Physical and Mathematical Sciences at Nanyang Technological University (NTU) in 2026, after completing a dual bachelor's degree at Beijing University of Posts and Telecommunications (BUPT) and Queen Mary University of London (QMUL).
I am also a core maintainer of Harbor, working with the Terminal-Bench team on scalable agent evaluation. I love open source and believe in its power to move research forward.
I am looking for PhD, RA, or research-internship opportunities, and I am open to research collaborations. Please feel free to reach out, even just for a coffee chat.
My research currently centers on the following topics:
I am fortunate to collaborate with Prof. Ludwig Schmidt, Lin Shi, Haowei Lin, Xiaoyue Zhou, and Xiang Li on agent benchmarking, and with Senkang Hu on self-evolving agents.
Our paper "Harbor Adapters and Harbor-Index: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation" was accepted to the NeurIPS 2026 Evaluations and Datasets Track.
Excited to join Tencent as an intern with the Hunyuan AI Data Department, working on frontier benchmark research.
Our paper "Unified Context Evolution for LLM Agents" was accepted to the EMNLP 2026 Main Conference.
New blog post: "Introducing Harbor-Index", a compact benchmark for agentic evaluation.
A gradient-free framework that externalizes agent experience into an evolving library of typed, reusable context units, retrieved at decision time and pruned when no longer valuable.
A unified adapter infrastructure for more than 80 agentic benchmarks, paired with large-scale model and harness evaluation and Harbor-Index, a curated 82-task meta-dataset spanning 29 benchmarks.
A compact, diverse, challenging, and high-quality benchmark for agentic evaluation.