Ph.D Candidate @ NUS Math
Ph.D Candidate @ NUS MathCurriculum VitaeEducationSelected Project ExperienceWorking ExperiencePublicationsConferences and TalksProject Dashboard
Contact
📞 Tel: +65-88299001
💬 WeChat: 18962528568
Languages
🇨🇳 Mandarin
🇺🇸 English
Interests
🎨 Drawing
🎼 Composing Music
🧗♂️ Climbing
About Me
I am a PhD candidate at the National University of Singapore (NUS), working at the intersection of AI and science. My research focuses on developing deep learning algorithms for complex dynamical systems, integrating physical insights to improve both accuracy and efficiency.
My journey in algorithm design began in 2021 during my internship at Heng Li’s Lab at Harvard Medical School, where I developed the Graph Waterfront Algorithm—now a core component of minigraph, a useful tool for sequence-to-graph mapper and graph constructor. This experience reinforced my passion for creating algorithms that are both theoretically sound and practically impactful.
At NUS, under the supervision of Prof. Qianxiao Li, I have been exploring AI for Science, particularly in high-dimensional dynamical systems. My work has led to sub-1% error models for solving complex physical problems and has been published in top computational journals, including Physica D and International Journal for Numerical Methods in Engineering.
I am actively seeking internship opportunities in computer vision (CV), large language models (LLMs), and AI for Science (AI4Sci). My broader interests include biological systems, physics, and robotics, where I aim to develop AI solutions that bridge deep learning and first-principles models. I would love to collaborate on impactful projects at the intersection of AI and science.
Top Skills
Machine Learning/Deep Learning Research • Algorithm Research • Python/C/C++ Programming • High Performance Computing • Problem Solving
Curriculum Vitae
- ShiqiWu_CV_2025_English.pdf [link]
- ShiqiWu_CV_2025_Chinese.pdf [link]
Education
Ph.D in Mathematics
National University of Singapore, Singapore
Aug. 2022 - July. 2026
Supervisor: Prof. Li Qianxiao (https://blog.nus.edu.sg/qianxiaoli/)
Research Interest: AI for Science, Computational Machine Learning
GPA: 4.83/5.00
Courses:
- Algorithms at Scale
- Computational Mathematics
- Modeling and Numerical Simulation
- Stochastic Analysis in Mathematical Finance
- Intelligent Robotics(Audit)
- Deep Learning and its Applications
- Advanced Mathematical Programming
- Linear Systems
- Game Theory and Applications
- Topics in Applied Mathematics
Selected Project Experience
National University of Singapore
Learning Dynamics of Nonlinear Field-Circuit Coupled Problems
Graduate Researcher, Supervised by Prof. Li Qianxiao & Prof. Ludovic Chamoin
Dec. 2023 - Dec. 2024
- Developed a hybrid model combining first-principles physics and machine learning, achieving 1% prediction error on a 7000-dimensional nonlinear field-circuit system.
- Designed and implemented a Non-intrusive Model Combination algorithm, fusing a physics-based state-space model with a Koopman-type deep neural network, enabling a 1000x speedup over traditional methods.
- Engineered a transformer-like encoder for Koopman-based dictionary learning, enhancing high-dimensional system representation.
- Led the full research pipeline, including dimensionality reduction, model construction, algorithm optimisation, experiment validation, and data analysis. Authored and published the research in International Journal for Numerical Methods in Engineering (JCR Q1, top-tier CFD journal), collaborating with co-authors for revisions and submission.
National University of Singapore
Non-intrusive Model Combination for Learning Dynamical Systems
Graduate Researcher, Supervised by Prof. Li Qianxiao
Dec. 2022 - Oct. 2023
- Developed a novel non-intrusive algorithm that seamlessly integrates physics-based and machine learning models, providing a unified framework for model combination.
- Proved the algorithm’s linear convergence under specific assumptions and incorporated efficient acceleration techniques to enhance performance across various machine learning tasks.
- Designed and implemented experiments using ResNet-based architectures, demonstrating applicability in robotics control and cardiac electrophysiology modeling.
- Led the research pipeline, including algorithm design, mathematical proof, and experimental validation. Authored and published the research in Physica D: Nonlinear Phenomena (JCR Q1, leading journal in nonlinear phenomena).
Harvard University & Dana-Farber Cancer Institute
Graph Wavefront Algorithm: Fast sequence-to-graph alignment algorithm
Mar. 2021 - Jun. 2022
- Proposed and developed the core concept of the Gwfa algorithm, enabling sequence-to-graph alignment with up to 10,000x speedup over existing exact algorithms.
- Implemented the initial C-based prototype, designing efficient data structures for storage and retrieval, laying the groundwork for large-scale validation and pruning optimizations.
- Conducted small-scale experimental verification, demonstrating the feasibility and accuracy of the algorithm.
- Developed the foundational exact algorithm, which was later integrated as the core algorithm in MiniGraph, a widely recognized tool in bioinformatics with 400+ GitHub stars.
Working Experience
National University of Singapore
Part-Time Teaching Assistant
ST5188:Advanced Data Science Project, Statistics Graduate Course
Jan. 2025 - Apr. 2025
- Mentor total 48(8 teams) postgraduates in Statistics data science projects, covering time series analysis, natural language processing (NLP), and computer vision.
- Provide technical guidance, helping students develop methodologies, refine models, and correct technical misconceptions. Reviewed their implementations, identified and clarified technical misconceptions, and guided them in troubleshooting coding issues.
- Facilitate effective communication between students and faculty, ensuring project alignment and research clarity.
Publications
- Wu, Shiqi, Gerard Meunier, Olivier Chadebec, Qianxiao Li, and Ludovic Chamoin. “Learning Dynamics of Nonlinear Field-Circuit Coupled Problems with a Physics-Data Combined Model.” International Journal for Numerical Methods in Engineering(2025). Accepted for publication.
- Wu, Shiqi, Ludovic Chamoin, and Qianxiao Li. “Non-intrusive model combination for learning dynamical systems.” Physica D: Nonlinear Phenomena(2024). https://www.sciencedirect.com/science/article/abs/pii/S0167278924001039
- Zhang, Haowen, Shiqi Wu, Srinivas Aluru, and Heng Li. “Fast sequence to graph alignment using the graph wavefront algorithm.” arXiv preprint arXiv:2206.13574 (2022). https://arxiv.org/abs/2206.13574
Conferences and Talks
- 14th AIMS Conference - Dec 2024 - Contributed Talk: “Non-intrusive model combination for learning dynamical systems.”
- AI for Science and Nobel Turing Challenge Initiative (AI4Sci/NTCI) Conference - July 2024 - Poster Presentation: “Non-intrusive model combination for learning dynamical systems.”
