Hello, I'm
Hao (Flower) Xu
CS PhD student at UC San Diego, building machine learning systems for large-scale retrieval, recommendation, and scientific discovery. Research spans foundation models, graph neural networks, and interpretable learning.
About
A Bit About Me
I am a CS PhD student at the University of California, San Diego, working under the supervision of Prof. Nuno Bandeira. My research focuses on developing machine learning methods — including retrieval, reranking, generative models, and graph neural networks — with applications in mass spectrometry-based proteomics and recommendation systems.
Previously, I was a Machine Learning Scientist Intern at TikTok Inc. (2025), designing graph neural networks for large-scale social recommendation, mentored by Dr. Tianxiang Tan (now at xAI). Before that, I worked as a researcher applying ML to computational biology at Queen's University with Prof. Laurence Yang, and as a machine learning consultant at Samumed LLC with Dr. James Yurkovich.
I hold an honoured MSc in Computer Science from the University of Sheffield, where I earned the Fretwell-Downing Prize for the best M.Sc. Dissertation under the supervision of Prof. Haiping Lu. I also received the D. E. Shaw Graduate and Postdoctoral Fellowship in 2023. Prior to that, I completed my undergraduate studies at Central South University in China.
Research
Research Interests
My research develops machine learning methods — foundation models, graph neural networks, and interpretable learning — applied to information retrieval, recommendation systems, and computational biology.
Retrieval & Ranking
Foundation models for large-scale retrieval and reranking, with applications in spectral library search, database search, and de novo sequencing.
Graph Neural Networks
Heterogeneous graph learning and link prediction on large-scale temporal networks for knowledge graphs, drug discovery, and social networks.
Recommendation Systems
Intention-enriched graph models for user behavior modeling and social recommendation on dynamic interaction networks.
Computational Biology
ML for proteomics, protein function prediction from 3D structures, and fragmentation ion probability benchmarking.
Interpretable ML
Explainable frameworks with sparsity learning to reveal mechanisms behind model predictions and adverse drug reactions.
Open Source
Tools & Software
PaperWritingLLM
Terminal AI writing assistant with fine-tuned Qwen2.5-7B for personalized writing feedback and corrections.
Pep2Prob
The first ML benchmark for fragment ion probability prediction. 183M spectra for 610K precursors.
PyKale
A PyTorch ecosystem library for knowledge-aware machine learning on multimodal data. 29k downloads.
APRILE
Explainable ML library for discovering adverse drug reaction mechanisms. 6k downloads.
TIP
Tri-graph information propagation for polypharmacy side effect prediction. NeurIPS GRL 2019.
GripNet
Graph information propagation on supergraph for heterogeneous graphs. Pattern Recognition 2023.
Updates
Recent News
Jun 2026
Poster: "Generalizable Supervised Learning for Spectral Library Searching Eliminates Per-Search Model Training" at ASMS Conference, San Diego.
Jun–Sep 2025
Machine Learning Scientist Intern at TikTok Inc., working on recommendation systems with graph neural networks.
Jun 2025
Poster: "Sensitive and interpretable supervised learning for peptide identification with spectral library search" at ASMS Conference, Baltimore.
Jun 2024
Oral presentation: "Benchmarking peptide spectral library search" at ASMS Conference, Anaheim.
May 2023
Received the D. E. Shaw Graduate and Postdoctoral Fellowship from D.E. Shaw Research.
Sep 2022
Started PhD at UC San Diego, advised by Prof. Nuno Bandeira.
Jul 2022
Presented at ISMB (MLCSB & 3DSIG COSIs), Madison, Wisconsin.
Dec 2019
Presented at NeurIPS Graph Representation Learning Workshop, Vancouver.