Hi, I'm Zihao GUO (Eric)
I'm a PhD student at Institut Polytechnique de Paris, investigating how to preserve graph performance under external perturbations, under the supervision of Dr. Andrea ARALDO and Dr. Mounim EL YACOUBI.
I hold a generalist engineering degree from École Centrale de Nantes, with a focus on applied mathematics. I enjoy collaborating on statistics and optimization.
Research interestsData driven decision makingOptimization under uncertaintyLLM post-training & self-improvement
Research
My research spans multiple threads focused on optimization under uncertainty, with an emphasis on large-scale decision-making in complex network systems.
- Formulating resilient optimization frameworks for vehicle routing and transport network design, prioritizing solution stability and service efficiency under multi-scale uncertainties.
- Advancing high-performance decomposition techniques, such as Column Generation, to tractably solve high-dimensional stochastic and combinatorial optimization problems.
- Integrating Deep Reinforcement Learning (DRL), Graph Neural Networks (GNNs), and Large Language Models (LLMs) to automate heuristic discovery and co-optimize real-time Learning-to-Optimize (L2O) pipelines for complex decision systems.
Publications
Full list on Google Scholar
|
Predict-then-Optimize Framework for Public Transport Line Redesign under Fluctuating Traffic Conditions [Link] Zihao Guo, Andrea Araldo, Faycal Touzout, Mounim El-Yacoubi Under review |
|
Data Sampling-driven Adaptive Modification of Bus Routes Under Time-Varying Road Conditions [Link] Zihao Guo, Andrea Araldo, Mounîm A El Yacoubi The 19th Learning and Intelligent Optimization Conference, 2025 |
|
Restoring Accessibility During Urban Rail Disruptions via Bus Network Redesign [Link] Zihao Guo, Andrea Araldo, Mounîm A El Yacoubi Transportation Research Board (TRB) 104th Annual Meeting, 2025 Transportation Research Record |
|
Rate variation and recurrent sequence errors in pandemic-scale phylogenetics [Link] Nicola De Maio, Myrthe Willemsen, Zihao Guo, Abhratanu Saha, Martin Hunt, Nhan Ly-Trong, Bui Quang Minh, Zamin Iqbal, Nick Goldman Nature Methods |
|
A Sinh–Cosh-Enhanced DBO Algorithm Applied to Global Optimization Problems [Link] Xiong Wang, Yaxin Wei, Zihao Guo, Jihong Wang, Hui Yu, Bin Hu Biomimetics |
|
SA-CNN: Application to text categorization issues using simulated annealing-based convolutional neural network optimization [Link] Zihao Guo, Yueying Cao ACM EITCE 2022. |
Education
Nantes Université
Master of Science degree · 2022 — 2023
Specialization in Mathematics and Applications.
Ecole centrale de Nantes
Engineer's degree - Generalist Engineer · 2021 — 2023
Focus on Applied Mathematics (Statistics & Data Science) and Research & Development.
Industrial Experience
Algorithm Engineer 2025.8 — 2025.11
Internship
@ Huawei — Shanghai, CHINA
Algorithm Developer 2023.4 — 2023.9
Internship
@ European Bioinformatics Institute — Cambridge, UK
Data Scientist 2022.4 — 2022.9
Internship
@ IFP Energies nouvelles — Rueil-Malmaison, FRANCE
Data Scientist 2021.11 — 2022.1
Internship
@ Tencent — Beijing, China (Remote)
Teaching
MSc - Intelligence artificielle et données
Teaching Assistant @ ENSTA Paris — Palaiseau, FRANCE
Professional Machine Learning Engineer Certification
Google Cloud · 2026.8 — 2028.8
Related roles
- Machine Learning Engineer: Build and maintain machine learning models and systems.
- Machine Learning Researcher: Research and prototype new machine learning methods.
- AI Engineer: Develop AI applications using software and hardware platforms.
Languages
Chinese: Native language
English: Fluent (GRE 326)
French: Fluent (DALF C1)
Research Scientist
ContractOct 2023 - Oct 2026
- Developed LLM-based systems for algorithm generation and optimization using SFT and RLFT.
- Integrated traffic forecasting with multi-objective optimization to enable proactive public transport network redesign under uncertainty.
- Designed vehicle routing methods under uncertainty with MILP and stochastic optimization.
Shenyang University of Aeronautics and Astronautics
Bachelor of Engineering · 2017 — 2021
Specialization in Mechanical Engineering.
Algorithm Engineer — Huawei
2025.8 — 2025.11 · Shanghai, CHINA
- Developed a multimodal contrastive learning framework integrating 3D point clouds and wireless channel data to predict channel characteristics at target locations from surrounding environmental observations, aligning both modalities in a shared representation space and validating predictions against ground-truth channel measurements.
- Proposed a non-contiguous resource block allocation algorithm based on integer partitioning, extending the conventional RIV-based contiguous allocation mechanism to support more flexible spectrum resource allocation.
Data Scientist — IFP Energies nouvelles
2022.4 — 2022.9 · Rueil-Malmaison, FRANCE
- Analyzed large-scale public transport validation data stored and queried in MongoDB, and developed a dashboard to monitor mobility patterns and key indicators.
- Compared observed validation data with MATSim agent-based transport simulations to identify discrepancies and assess simulation accuracy.
- Developed time-series forecasting models (SARIMAX, NeuralProphet) to analyze influencing factors, predict passenger flows, and correct statistical estimates.
Supervised by Dr. Pierre Michel.
Algorithm Developer — EMBL-EBI
2023.4 — 2023.9 · Cambridge, UK
- Developed and evaluated large-scale phylogenetic algorithms for simulating genome evolution along large, short-branch phylogenetic trees, supporting datasets of up to 500,000 genomes.
- Built an automated Python/R/Bash pipeline for simulation, cluster execution, result collection, and benchmarking across multiple experimental configurations.
- Evaluated topological accuracy, likelihood, runtime, and memory consumption over repeated experiments, achieving a >100× speedup compared with the reference method while maintaining competitive phylogenetic accuracy.
Supervised by Dr. Nicola De Maio.
Data Scientist — Tencent
2021.11 — 2022.1 · Beijing, China (Remote)
- Developed a customer churn prediction model on approximately 100,000 customer records, engineering 30+ behavioral and usage features to identify high-risk users and support targeted retention strategies.
- Trained and compared 4 classification models using scikit-learn, optimized them with GridSearchCV, and achieved an ROC-AUC of approximately 0.86, with the top 20% highest-risk customers capturing around 55% of observed churners.
Supervised by Dr. Zhao Zhang.