I am a Transportation PhD candidate at Beijing Jiaotong University.
我关注如何把真实物流场景转化为 Agent 能理解、算法能约束、人能验证 的决策流程。
Personal Website · Jiaoda Weilan · Freight Matching Agent · my-quant-lab
Scenario → Constraint → Agent
- Scenario: freight-searching, campus delivery, vehicle routing, time windows, user preferences, and operational bottlenecks.
- Constraint: translate business rules, uncertainty, capacity, energy, risk, and natural-language preferences into models, rules, scoring, and validation gates.
- Agent: use LLM workflows, prompt engineering, tool use, state tracking, and full-stack demos to make decision support interactive and inspectable.
- Logistics AI Agents that combine LLM understanding with optimization-aware decision logic.
- Vibe-coding driven full-stack demos, with tests, documentation, bad-case review, and public-safe project packaging.
- Research and prototypes around VRP / EVRPTW / pickup-and-delivery routing, robust optimization, and intelligent logistics decisions.
| Project | What to look for |
|---|---|
| 🎓 Jiaoda Weilan / 交大微澜 | Conversational campus delivery Agent, product thinking, full-stack demo, trusted LLM + backend boundary |
| 🚚 Freight Matching Agent | Agentic AI prototype for continuous freight-searching; Top 24 / 963, finalist award |
| 📈 my-quant-lab | Python research workbench for validation discipline, backtesting, Streamlit, and AI-assisted engineering |
- ASCE Journal of Urban Planning and Development: accepted / in production.
- ICTTS 2026 EI conference paper: accepted.
- Additional SCI manuscripts are under review and described conservatively until formal acceptance.
- Graduate National Scholarship; master's ranking 14 / 152.
Python · MATLAB · Java · TypeScript/JavaScript · Git
Prompt Engineering · AI Coding · Agent Workflow · Bad-case Review
VRP · MILP · Heuristic Search · SAA · Wasserstein DRO · CVaR
This profile only contains public-safe materials. It does not publish my private resume, phone number, private email, recruiter conversations, credentials, screenshots with restricted data, or application records.
Building logistics decisions that an Agent can explain, an optimizer can constrain, and a human can verify.