BIM RAG
Research-driven BIM retrieval combining validated SQL, semantic RAG, and IFC graph search to answer natural-language questions and highlight grounded results in a 3D viewer.
Design Technologist AL/ML Engineer Computational Designer
I build AI and computational tools for buildings, cities, and spatial systems.
I work where architecture, engineering, and construction meet AI, machine learning,
geospatial analysis, and data-driven design. I turn spatial problems into systems,
tools, and models that inform how we plan, design, and build.
01
AI & Machine Learning
Retrieval systems, computer vision, and LLM-driven workflows applied to design and spatial data.
02
Computational Design
Parametric modeling, generative systems, and automation across the design and construction process.
03
Geospatial & Data
Spatial analysis, mapping, and data pipelines that make urban and building data legible.
Selected workCurated
Neural Floorplan
A deep-learning pipeline that converts raster floorplans into classified vector CAD via SegFormer semantic segmentation, using frozen-backbone transfer learning on CubiCasa5K.
Explorentory
An AI-powered NYC rental discovery platform that replaces rigid filtering with trade-off exploration, combining machine-learning personalization and conversational LLM explanations across 4 million synthetic property records.
Street-Block Urbanity Prediction
Using linear regression, logistic regression, and a graph neural network, this project tests how much of a city's character can be predicted from street-block geometry alone.
Manhattan Transit Accessibility Mapping
Built in Rhino Grasshopper and C#, this tool calculates and maps the walking time from every building in Manhattan to its nearest subway station through shortest-path analysis of the pedestrian network.
GeoEstateChat
An exploration of how LLMs can connect real estate with multi-scale geospatial analysis, mediating between spatial data, analytical workflows, and real-estate reasoning through user intent.
Parametric Study of Hong Kong Cruciform Towers
An architectural design project that redefines Hong Kong's cruciform tower typology through in-depth parametric analysis of building codes, design grammar, and view corridors.
Spiral Dwelling
An architectural design project reimagining Hong Kong's podium tower through generative iterations, exploring spiral geometries and walkability analysis.
Parametric City Generation & Analysis
A parametric city generated in Grasshopper from street networks, building geometry, and density inputs to produce adaptable urban forms, paired with solar and view analyses.
No True North
An interactive cartographic tool that redefines the Earth's north pole to generate alternative world maps, revealing how projection geometry, seam placement, and distortion reshape how the world is represented.
Bryant Park Solar Mapping
A parametric analysis of one year of solar radiation in Bryant Park, modeled with Grasshopper, Ladybug, and the DeCodingSpaces toolbox.
Manhattan Residential Clustering
K-means clustering that groups Manhattan's residential buildings into interpretable themes, drawing on eleven residential datasets.
Mergeprep
A Python package for reliable row-wise merging of messy tables. It prepares tables for merging, finds matching columns, suggests merge keys, and explains why merges succeed or fail.