Publications
Journal articles, conference papers, and workshop publications in urban environmental science, AI, and human-computer interaction.
First-author publications
2026
- JWEIA
Design trade-offs in building porosity: A parametric analysis of vertical placement and geometry for urban ventilationJournal of Wind Engineering and Industrial Aerodynamics, 2026A validated RANS parametric study shows that the vertical location of a porous opening matters more than its size: ground-level porosity preserves pedestrian wind, while elevated voids can suppress it.
- IAQVEC
Aerodynamic response and ventilation enhancement in urban street canyons through building porosity: A comparative numerical and experimental studyIn E3S Web of Conferences, 2026Wind-tunnel measurements and validated RANS simulations compare three void-deck geometries, showing that a slot reaching toward the ground is most effective at flushing the pedestrian zone.
2025
- NeurIPS
Physics-Informed Large Language Models for HVAC Anomaly Detection with Autonomous Rule GenerationIn NeurIPS 2025 Workshop on UrbanAI, 2025Selected oral contributionPILLM evolves interpretable HVAC anomaly-detection rules through physics-informed reflection and crossover, keeping AI-generated diagnostics grounded in thermodynamics and control logic.
- COBEE
The role of building porosity in enhancing passive ventilation in high-density urban street canyonsIn 6th International Conference on Building Energy and Environment, 2025A conference study of building porosity as a passive-ventilation strategy for improving airflow in high-density urban street canyons.
- ICML
BuildEvo: Designing Building Energy Consumption Forecasting Heuristics via LLM-driven EvolutionIn ICML 2025 Workshop, 2025Poster presentationBuildEvo uses LLMs within an evolutionary loop to generate transparent, physically grounded heuristics for building-energy forecasting.
2023
- Sci. Data
District-scale surface temperatures generated from high-resolution longitudinal thermal infrared imagesScientific Data, 2023A rooftop infrared observatory in Singapore produced 1.36 million district-scale thermal images, enabling fine-grained study of buildings, roads, and vegetation over time.
- Water Res.
Coagulant dosage determination using deep learning-based graph attention multivariate time series forecasting modelWater Research, 2023GAMTF combines graph attention with time-series forecasting to jointly predict coagulant dosage and settled-water turbidity from long-term treatment-plant data.
- JWPE
Comparing artificial and deep neural network models for prediction of coagulant amount and settled water turbidity: Lessons learned from big data in water treatment operationsJournal of Water Process Engineering, 2023Six years of operations data reveal why time-aware deep models improve predictions of both coagulant dosage and turbidity in a changing water-treatment process.
Co-authored publications
Papers on which I contributed as a co-author.2024
- JWPEDeep learning-based coagulant dosage prediction for extreme events leveraging large-scale dataJournal of Water Process Engineering, 2024
- Chemosphere
2021
- CHI EAHow the death-themed game Spiritfarer can help players cope with the loss of a loved oneIn CHI ’21 Extended Abstracts, 2021Equal contribution