Mingxue (Mercy) Xu
"Loyal to the Truth."
Hi there:D
I completed my PhD at Imperial College London in May 2026, supervised by Prof. Danilo P. Mandić. My research focuses on Generative Language Model Compression with Algebraic Approaches, including tensor decomposition, structural pruning, and deployment on resource-constrained devices (e.g. Raspberry Pi).
I am also the founder of Facthink, building Graph-Hypoth, a multi-agent system that develops research claims into candidate hypotheses and literature-grounded experiment plans.
I am currently on the job market, looking at the intersection of research and engineering. Please do keep in touch for any potential opportunities!
Previously, I obtained my BEng at China University of Petroleum - Beijing, and did my MEng at University of Science and Technology of China, under the supervision of Prof. Xiang-Yang Li. Both my BEng and MEng are in Computer Science and Technology. I also interned at Baidu PaddlePaddle as a technical product manager, serving for the developer community of PaddleSpeech.
news
| Oct 7, 2026 | Gave a talk at multi-agent security at Oxford Collaboration Week, taking Graph-Hypoth as a case study, slides are here. |
|---|---|
| Jul 31, 2026 | Surviving Resource-constraint Compression: Capability Retention under Tensor-train Decomposition for Sub-billion Parameter Language Models is published by Machine Intelligence Research. |
| Mar 5, 2025 | How can representation dimension dominate structurally pruned LLMs? is accepted by ICLR 2025 Workshop on Sparsity in LLMs (SLLM). |
| Oct 7, 2024 | Manuscript Geometry is All You Need: A Unified Taxonomy of Matrix and Tensor Factorization for Compression of Generative Language Models is available online. |
| Sep 27, 2024 | Co-authored paper, FinLlama: LLM-Based Financial Sentiment Analysis for Algorithmic Trading, has been accepted by the International Conference on AI in Finance (ICAIF2024). |
| Jul 4, 2023 | Manuscript TensorGPT: Efficient Compression of the Embedding Layer in LLMs based on the Tensor-Train Decomposition is available online. |
| Jul 2, 2023 | Private Training Set Inspection in MLaaS is featured by Montreal AI Ethics Institute, for the details please drop in here. |
| Mar 28, 2023 | Co-authored paper at Baidu, PaddlePaddle: A Production-Oriented Deep Learning Platform Facilitating the Competency of Enterprises, is available online. |
| Jan 29, 2023 | Manuscript Data Origin Inference in Machine Learning is available online (code included). |