Selected work across urban analytics, physical infrastructure and applied AI.
A self-contained coding agent you run from your terminal: a Python process hosts the harness (LangChain, tools, memory, optional Langfuse tracing), and an optional Ink + React TUI sends work over NDJSON on stdio so scripts or a UI can drive the same backend.
Digital twin of New York City's urban canopy that simulates tree shade on streets and buildings using LiDAR-based 3D tree models.
Python toolkit that extracted, enriched, and visualized NYC's urban forestry from billions of LIDAR points.
Regression analysis using demographic and neighborhood data to quantify gentrification patterns in NYC.
Brain-Computer Interface (BCI) powered exoskeleton that converts EEG signals into movement commands.
