Curriculum Vitae

Download Full CV (PDF)

🎓 Education

• Harbin Institute of Technology (HIT), China
  • Ph.D. Student (Incoming)
• Technical University of Munich Asia (TUM Asia), Singapore & Germany
  • Master of Science (M.Sc.) in Railway Engineering (ITS Focus) (Sep 2023 – Feb 2026)
  • Thesis: Simulation-Based Bidirectional Control between AVs and Adaptive Traffic Signal Control Using Real-World V2X-Infrastructure
  • Relevant Courses: Traffic Operation and Control (ITS), Transportation Modelling and Simulation Tools, Rolling Stock, Trackwork, Public Transport.
• Southwest Jiaotong University / University of Leeds (Joint Program), China & UK
  • Bachelor of Engineering (B.Eng.) in Civil Engineering with Transport (Sep 2019 – Jul 2023)
  • Thesis: Influence of Sasobit on Physical and Chemical Properties of Asphalt
  • Relevant Courses: Structural Design and Analysis, Railway Engineering, Highway Engineering, Transport Planning and Modelling.

💼 Work Experience

• AI Agent Developer

  Sihe Information Technology Co., Ltd., Harbin, China

  • Jun 2026 – Present
  • Designed and developed Long-horizon Task Agents and reusable Agent Skills (SOPs) for complex task decomposition, sustained execution, state management, and tool calling.
  • Researched fundamental LLM mechanisms (Tokenization, BPE) and model distillation (Teacher–Student, knowledge transfer, model lightweighting).
  • Drove rapid prototyping on complex projects using agent tools (Codex, Hermes, pi) for vibe coding, code generation, and automated workflows.

• L4 Autonomous Driving & ITS Engineering Intern

  IABG mbH, Munich, Germany

  • Jul 2025 – Feb 2026
  • Worked on the Mobility Innovation Campus (MIC) L4 autonomous driving test site, conducting R&D and engineering validation of autonomous driving and ITS.
  • Fused Camera, LiDAR, Radar, and V2X infrastructure information with Bird’s-Eye-View (BEV) scene representations for complex traffic scene understanding and intersection coordination.
  • Developed Python-based data processing and real-time traffic interaction/control logic (RSU/OBU, SPaT, V2X), supporting trajectory analysis, L4 system testing, and scenario validation.

• Traffic Engineer Intern

  ST Engineering Urban Solutions Ltd., Singapore

  • Aug 2024 – Feb 2025
  • Automated VISSIM simulation workflows in Python and built, calibrated, and optimized PTV VISUM/Optima models for traffic operations analysis and control optimization.
  • Integrated multi-source road network, GIS, detector, and OD data (QGIS, GeoPandas, Pandas, SQL) for ITS model optimization, traffic state analysis, and travel demand forecasting.

🔬 Research Experience & Projects

• Toward Continually Growing World Models: An OaK-Inspired Architecture for Learned Abstraction and Planning (Aug 2026)
  • Formulated Structural Continual World Modeling (SCWM), integrating parameter-level adaptation ($\theta_t \to \theta_{t+1}$) with a structural adaptation axis ($\mathcal{K}t \to \mathcal{K}{t+1}$) based on Richard Sutton’s OaK architecture and FC-STOMP lifecycle.
  • Formalized predictive state representation expansion with constructed features, reward-respecting subtasks with terminal bonuses, predictive option models, and Pareto-optimal utility pruning.

  • Published preprint with Zenodo DOI: 10.5281/zenodo.22103257

    GitHub: AnXu-ITS/OaK-World-Model.

• From Foresight to Reactive Traffic Worlds: A Survey of Driving World Models for Closed-Loop Simulation (Aug 2026)
  • Authored comprehensive survey investigating Driving World Models as learned closed-loop simulators, focusing on counterfactual off-log environmental feedback and multi-agent reactivity.
  • Formulated a 3-regime Simulation-Closure framework (Replay-Constrained $\rightarrow$ Ego-Responsive $\rightarrow$ Traffic-Reactive) and established a 5-layer claim-specific validity stack spanning observation fidelity, ego controllability, traffic-response validity, policy validity, and corridor/network scale.

  • Released complete preprint on SSRN: SSRN: 7383662 (DOI: 10.2139/ssrn.7383662), reproducible LaTeX package, and editable PowerPoint figures. GitHub: AnXu-ITS/driving-world-models-survey

    PDF: Download Full Survey.

• Conditional Effects of Connected-Vehicle Requests in Reinforcement-Learning Traffic Signal Control (Jun 2026)
  • Disentangled queue-responsive RL adaptation from connected-vehicle request conditioning through matched DQN ablations across 2,730 one-hour SUMO evaluation runs.
  • Conducted 2-dimensional Crossed-Bootstrap uncertainty quantification (50,000 resamples) across 5 training seeds $\times$ 10 paired traffic seeds over 3 demand levels and 7 CV penetrations.

  • Released open-source benchmark codebase, pre-trained weights, and full evaluation logs. GitHub: AnXu-ITS/cv-request-dqn-tsc

    PDF: Download Manuscript (PDF).

• Simulation-Based Bidirectional Control between AVs and Adaptive Traffic Signal Control Using Real-World V2X-Infrastructure (Aug 2025 – Jan 2026)
  • Master’s Thesis / IABG-MIC Project
  • Built a bidirectional AV–signal control coordination framework on MIC’s real-world V2X infrastructure using a SUMO/TraCI microscopic simulation environment.
  • Designed a DQN-based adaptive traffic signal control algorithm for dynamic phase optimization in mixed traffic, evaluating queue length, delay, efficiency, and robustness under varying V2X penetration rates.
• AI-Driven Transformation of Modern Transportation Systems (TUM Asia Workshop)
  • Analyzed AI’s transformative impact across ITS, autonomous driving, smart railways, and urban traffic management.
  • Authored the paper AI’s Transformation of Transportation in Modern Society.
• Influence of Sasobit on Physical and Chemical Properties of Asphalt (Bachelor’s Thesis, Oct 2022 – Apr 2023)
  • Experimental testing and performance evaluation of the Sasobit additive on the physical and chemical properties of asphalt mixtures.

🛠️ Skills & Expertise

• AI Agent Engineering: Long-horizon Task Agents, Agent Skill & SOP Design, Tool Use, Workflow Orchestration, Vibe Coding (Codex, Hermes, pi)
• AI/LLM & Algorithms: Continual Learning, World Models, OaK Architecture, Reinforcement Learning (DQN, Options), Model Distillation, Tokenization/BPE
• Programming & Data Science: Python, SQL, MATLAB, Git, Pandas, NumPy, GeoPandas, Matplotlib
• Autonomous Driving & ITS: V2X (RSU/OBU, SPaT), Sensor Fusion (LiDAR, Camera, Radar), BEV Representation, Trajectory Analysis, SUMO, TraCI, PTV VISSIM, PTV VISUM, PTV Optima
• Engineering Tools: QGIS, AutoCAD, ANSYS, LaTeX, Markdown

🌐 Languages

• Chinese: Native
• English: Fluent

👥 Referees

• Dr. Martin Margreiter — IABG Industrieanlagen-Betriebsgesellschaft mbH, Munich, Germany (Email: margreiter@iabg.de)
• Liu Xiaodong — (Email: xiaodong.liu@outlook.com)