Artificial Intelligence & ICT · UNU Macau
I research agentic AI, build the systems that run it, and turn both into policy for the United Nations.
Applied AI researcher and institutional technologist at the United Nations University. I work across three fronts at once: peer-reviewed research on agentic AI and computational social science, the production cloud platforms that deliver it inside a UN institute, and the policy work that carries it to governments.
Agentic AI, LLM-enhanced agent-based modeling, computer vision, and remote sensing, published in Q1 journals and at IEEE and AAAI venues.
Sole ICT focal point at UNU Macau: I design, deploy, and operate the institute’s AI platforms and cloud infrastructure end to end.
Research that reaches decision-makers: AI as a digital public good, open-source readiness, and capacity building for the UN and member states.
Selected work
Agentic AI · Geospatial
LLM-driven agents that simulate a city, block by block.
A reproducibility-first framework for LLM-enhanced agent-based urban simulation. Spatially grounded agents perceive, plan, and act inside a minute-resolution model of Macau to test policy under heat, access, and misinformation.
Remote sensing · Open science
A large-scale remote-sensing road dataset, released openly.
A large, fine-grained remote-sensing road dataset paired with a vision-language-geometry algorithm for hierarchical road classification, released openly with a DOI for the research community.
Geospatial · Environmental health
Turning live environmental data into personal health guidance.
A geospatial platform that fuses IoT, GPS, mobile-sensing, and meteorological data into location-based health-risk assessment and personalized, hazard-aware navigation. The research behind it reached two top-tier Q1 journals.
Systems · Cloud · DevOps
The full AI and cloud stack of a UN institute, run by one person.
As the only ICT focal point at UNU Macau, I designed and operate the institute’s AI platform end to end: chat and agent services, an LLM router, authenticated tool portals, and the multi-cloud and on-prem infrastructure under them.
AI for education · Platform
AI-augmented Model UN that turns youth debate into policy insight.
An AI-enhanced Model UN initiative, with the Learning Planet Institute, that uses agentic AI to support delegates, amplify under-represented perspectives, and synthesize youth deliberation into structured reports.
Agentic systems · Research infrastructure
A reproducible agent harness for large-scale content analysis.
An end-to-end agent framework that brings modern engineering to qualitative research: a YAML recipe language, content-addressed caching for byte-level reproducibility, and claim-level evidence traceability.
LLM evaluation · with IBM Research
An evaluation framework comparing LLM policy recommendations with expert judgment across four cities on four continents. Best Paper, AAAI 2026 workshop.
Paper (arXiv)Capacity building · Microsoft-funded
Co-developed a global online course translating AI research into applied learning for diplomats, development practitioners, and policy analysts.
AI MacauComputer vision · IEEE IJCNN 2024
Detecting AI-generated images and the universal artifacts shared across diffusion models and GANs, with analysis of adversarial robustness.
Web · Cloudflare edge
A public reference app on Cloudflare Workers, D1, and R2 (React, Vite, TypeScript), showing cost-efficient edge deployment.
trip.jiaan.meInternship · ITU Geneva
Evaluated, integrated, and deployed an editor application on AWS to automate the ITU Operational Bulletin, with internal documentation.
GitHubPolicy research · ADB-funded
Defining and operationalizing open, governed AI as public infrastructure for development contexts across the Asia-Pacific.
Peer-reviewed research across agentic AI, remote sensing, and environmental science, including two Q1 journals and a Best Paper award.
What would an LLM do? Evaluating policymaking capabilities of large language models
CoRR abs/2509.03827 · Best Paper, AAAI 2026 Workshop on LMs for Underserved Communities
Emerging Media, 3(3), 401–413
Cognitive urban simulation with large language models under the fifteen-minute-city paradigm
In AI for Humanity: Building an Equitable Digital Future. Springer (UNU–Springer)
arXiv:2603.21222 · dataset on Zenodo (DOI 10.5281/zenodo.18642747)
Unveiling universal forensics of diffusion models with adversarial perturbations
IEEE International Joint Conference on Neural Networks (IJCNN)
Urban Climate, 40, 101020
Speaking & engagement
Research Assistant & ICT Focal Point (Artificial Intelligence) · Macau SAR, China
Operational Bulletin Automation Intern, TSB · Geneva (remote)
Team Lead Intern, International Exchange & Judicial Cooperation · Shanghai, China
Team Lead Intern · Hangzhou, China
M.Eng., Computer Science & Technology (AI specialization)
GPA 3.8/4.0. Minor in International Organizations & International Development (3.92/4.0). Graduated with Honors; Outstanding Student Award.
B.Mgmt., E-Commerce Technology
Interdisciplinary training across computer science, management, economics, and law.
Get in touch