MEMENTO is a real-time platform that captures urban experiences happening across the city, aiming to prevent people from drifting into oblivion (the state of being unaware or unconscious of what is happening) and instead transforming those moments into engaging odysseys (a long and eventful or adventurous journey or experience).
It empowers users to discover, capture, and engage with the mementos around them during their commutes, turning everyday journeys into moments of exploration, creation, and interaction. MEMENTO serves as an interaction, intersection, and interplay between the city, its people, and their experiences — a platform: By the people and the city, For the people and the city, Of the people and the city.

The Machine Learning pipeline for public mementos in MEMENTO is designed to automate the classification and tagging of public content sourced from platforms such as Secret NYC, Reddit, NYC Bucket List, Newsbreak, and more.
The pipeline consists of three primary models — Category Classification, Tag Prediction, and Duration Estimation — each trained using supervised learning techniques. Logistic Regression with One-vs-Rest strategy and Random Forest classifiers are employed to predict relevant categories and multiple tags, while Decision Trees handle ordinal duration estimation. By implementing this ML pipeline, MEMENTO enhances the consistency and accuracy of content classification, enabling the platform to transform unstructured public content into structured, contextually enriched mementos that align with its existing taxonomy.

Studio-In-Flux is an experimental research project that examines how spatial configurations influence movement and interaction within the Columbia GSAPP Studio. By analyzing the dynamic interplay between people, objects, and spatial arrangements, the project seeks to uncover hidden circulation patterns and behavioral trends that shape the studio environment. The goal is to reimagine workspace design by making these invisible dynamics visible and actionable.

The project employs a computational simulation built on agent-based modeling, mapping the studio as a responsive digital environment. The simulation tracks how agents (students, faculty, staff) navigate desks, chairs, and pathways, generating real-time heatmaps and efficiency scores for different spatial layouts. Using data-driven analysis and spatial AI, Studio-In-Flux identifies bottlenecks, optimizes circulation, and explores how spatial interventions can foster more effective, engaging studio experiences.

This research delves into the disconnection between people and their urban surroundings, examining how gaze patterns, physical interactions, and time-location dynamics have transformed commutes into passive, oblivious routines. By mapping these behaviors in Oblivion, the study highlights the erosion of meaningful engagement with the city.

In response, Odyssey emerges as a speculative intervention, reimagining commutes as opportunities for active exploration and connection. Designed for those seeking more than routine transit, it leverages AI-driven personalization and spatial computing to turn city streets into interactive playgrounds, fostering deeper interactions and rediscovery of the urban landscape.

BENTO is a next-generation platform designed to optimize retail store management through data-driven spatial analysis and algorithmic geometry. The project leverages customer behavior insights, spatial flow patterns, and sales data to generate adaptive store layouts that enhance both product placement and operational efficiency. By transforming retail spaces into dynamic, data-responsive environments, BENTO aims to improve customer experiences while maximizing space utilization.

To achieve this, BENTO integrates computational design tools like Rhino and Grasshopper, utilizing an Adjacency Layout Generator to create context-specific spatial configurations. The platform incorporates real-time data inputs from NYC Open Data, enabling tailored analysis for specific urban settings. Machine learning algorithms further refine layout strategies, generating actionable recommendations that align spatial arrangements with observed user behavior. Through this data-driven approach, BENTO bridges the gap between spatial design and retail management, creating more intuitive, impactful retail environments.

Hue Knew is an interactive quiz game that challenges players to identify films based on blurred stills and posters, testing their ability to deduce genre, release year, and title from abstracted visuals. By integrating the TMDB API, the game ensures a continuously updated content pool, creating a dynamic and engaging cinematic experience.

Data visualization plays a key role in Hue Knew, with a dashboard that tracks player performance, revealing insights into film recognition patterns and genre-specific trends. This analytical layer transforms the game into a learning tool, merging entertainment with data-driven exploration of cinema.

In a world where screens dominate commutes, Gaze Tracker seeks to rewire attention patterns by tracking gaze directions within a 3x3 visual grid. By mapping head positions and logging time spent in each zone, it reveals habitual downward gazing and encourages users to redirect focus outward, fostering a more intentional engagement with urban surroundings.

Combining real-time feedback with spatial analysis, Gaze Tracker reframes attention as both a trainable muscle and a moldable practice. Each notification serves as a nudge to look up, reconnecting commuters with the city and transforming routine commutes into opportunities for spatial awareness and discovery.

The Urban Fringe explores the spatial thresholds where public sidewalks intersect with private eating establishments, examining how these in-between zones shape urban experiences. These transitional spaces act as vital connectors, influencing social interactions, pedestrian movement, and the overall vibrancy of city life. By focusing on the interplay between public and private realms, the project reveals how these urban fringes serve as dynamic stages for everyday encounters and social behaviors.

To investigate these zones, the project employs computational analysis and spatial mapping to capture and analyze patterns of movement, social interaction, and spatial usage. Using data-driven mapping systems and digital tools, it visualizes how people navigate these thresholds, identifying key factors that impact their use and influence. The study ultimately provides insights into how urban fringes can be reimagined to foster greater connectivity, interaction, and vitality within the city.

Building upon the Urban Fringe project, Sidewalk Reclamation delves deeper into the spatial dynamics of sidewalk encroachment, focusing on radiation exposure and claustrophobic zones. By analyzing how environmental factors impact pedestrian experiences, the project identifies key areas where natural interventions, such as green buffers and shade structures, can mitigate heat and enhance walkability.

Leveraging computational modeling, the project examines radiation patterns, sunlight distribution, and spatial compression across sidewalk zones. Data-driven simulations inform targeted design solutions, integrating natural elements to not only reduce heat exposure but also create more welcoming and accessible public pathways.