
Automating Topology Corrections for Fragmented Networks
2025 – 2026 · llm · spatial-ml
An LLM-based interpretable model for correcting fragmented network structures in pedestrian, cycling, and utility infrastructure.
Research and applied projects spanning vision-language modeling, spatial ML, LLM reasoning, urban computing, and design.

2025 – 2026 · llm · spatial-ml
An LLM-based interpretable model for correcting fragmented network structures in pedestrian, cycling, and utility infrastructure.

2026 · llm
An LLM pipeline that generates hierarchical, sentiment-aware product-review taxonomies — Wayfair × MIT 15.S04 Generative AI Lab.
2025 – 2026 · spatial-ml · urban-computing
Data-driven pedestrian volume modeling for infrastructure and safety planning across Maine.

2025 · vision-language
Analyzing temporally averaged visual representations for alignment with human engagement signals — alternative to standard VLMs.

2025 · urban-computing
Freedom Summer Fellowship 2025 — data-driven cool roof program for climate resilience.

2025 · llm · urban-computing
Generative AI and MLOps for automated water management, anomaly detection, and predictive maintenance.

2025 · ui-ux · vision-language
A multi-modal way to learn Chinese characters via voice + visual interaction.
2024 · spatial-ml · urban-computing
Analyzing pedestrian behavior and infrastructure in Boston using spatial data.

2024 · urban-computing
Evaluating EV infrastructure accessibility in disadvantaged communities.

2024 · spatial-ml
Deep learning approach for processing sparse matrices directly without low-rank approximations.

2023 · spatial-ml · urban-computing
Optimizing wildfire evacuations through integrated traffic simulation strategies.