Data

This project builds an end-to-end natural language interface over the NYPD Civilian Complaint Review Board (CCRB) complaints dataset. Using DuckDB and a large language model, users can query structured complaint records, covering officer demographics, allegation types, dispositions, and penalty recommendations, using plain English questions.

This project supports KAVI (Kings Against Violence Initiative) by mapping census tract-level demographic and economic trends, including poverty, employment, race, income, rent, and home prices, alongside NYPD precinct boundaries and geocoded shooting incidents from 2006–2024. The goal is to provide a data infrastructure for understanding how economic and demographic conditions relate to community violence over time.

This project focuses on youth detention rates in New York City using ACS detention and placement demographic reports, including ZIP code-level data from NYC Open Data and FY24 detention demographics. The work supports building a cleaned dataset and visual analyses that connect youth detention trends to neighborhood characteristics.

This project builds a Retrieval-Augmented Generation (RAG) pipeline on a collection of press releases published by New York District Attorneys regarding gang indictments. A RAG model is a model that is trained on specific documents and specializes in answering questions about those documents. The goal is to extract structured, queryable data from unstructured legal documents making it easier to analyze patterns in gang-related prosecutions across NYC boroughs. This notebook is a starting foundation for hackathon participants to explore and extend in any direction.

The project develops a data infrastructure linking New York City gender based violence (GBV) incidents with policing and mass incarceration related data. It integrates NYPD incident data, stop and frisk records, gun violence data, demographic variables, and spatial boundary files. The goal is to enable analysis at the city council district level despite current data scale and privacy limitations. By combining spatial disaggregation methods with research on local ecologies of GBV, the infrastructure aims to support policy accountability and inform community based, non police responses to GBV.

This project focuses on developing community defined measures of safety, justice, and wellbeing in the Bed-Stuy neighborhood of Brooklyn. Building on the Everyday Peace Indicators framework, the project uses participatory research methods to gather neighborhood level insights through surveys, interviews, and collaborative engagement. The resulting data supports the creation of indicators that reflect residents’ lived experiences and informs tools such as a neighborhood “pulse check” app to track community wellbeing and support community based approaches to safety.

Develop a Safe Cities-inspired platform for Edmonton, Alberta’s Blueprint for Violence Prevention. The platform would serve as a tool to help understand dispersed data sets from multiple sectors and identify the most pressing violence-related problems in Edmonton, their main determinants, and the evidence-based solutions needed to address these issues.

Complexo do Viradouro and Caramujo are areas in Niteroi that have long suffered high criminal activity and isolation. It has now been redeveloped for the general public, but the data analysis has been tricky. The project sponsor wants to know how to collect and analyze data that determines whether the redevelopment has actually caused less crime in the area, or if there is a "no snitching" code causing fewer crime reports.