Project highlights

Here's a sample of data projects I've built, edited, and/or directed.

120,000 people purged from Brooklyn voter rolls

Our WNYC Data News Team noticed that tens of thousands of voters were suddenly missing from the NYC voter rolls. Within days we found that the number was twice as large as initially believed, and that Hispanic voters had been disproportionately purged. Our ongoing reporting and analysis led to state and federal investigations, and eventually to a consent decree mandating the board of elections overhaul its procedures. And, importantly, our work led to the restoration of those voters to the rolls.

Mean Streets: Tracking deaths in NYC

As the murder rate dropped in New York City, residents became concerned with another killer: cars. Pedestrians, bicyclists, and others were dying at an alarming rate. But there was no firm data about exactly how many people were being killed. Several different departments disagreed on the stats, and nobody's numbers matched those of the police. So we started tracking every fatality, cross-referencing official figures, press, and social-media reports. Our became the de-facto "official" count as the city moved to implement "Vision Zero" — to bring traffic deaths to zero — and our coverage kept a close eye on that effort.

Superstorm Sandy: Deadly topology

This was a challenge: How to clearly illustrate a story about how a bowl-shaped neighborhood on Staten Island suffered an unusual concentration of deaths during Sandy — likely because they were caught off guard by water rushing into that basin. I worked from an open data set of elevation contours to design and build this map, experimenting with many approaches to the data and presentation until it was as clear as possible. The deployed version, which no longer functions, was a "slippy map" you could pan and zoom.

"Resisting arrest" in black and white

The WNYC Data News team I led, working with an investigative reporter at the station, revealed that NYPD officers were fare more likely to add "resisting arrest" charges for black suspects than for white suspects. Since we looked at the pool of people already arrested  for drug crimes, disorderly conduct, and theft over five years, this calculation suggested a bias beyond the systemic racial imbalance of overall arrests.

The cops behind NYPD "resisting arrest" arrests

While the illustration is simple, the work behind it was not at all. In that same series, we analyzed five years of court records to show that a small percentage of NYPD police officers were behind the majority of the "resisting arrest" charges, a charge known to be used to justify use of force.

A bot to monitor federal courts

I built a bot that watched federal court filings for specific criteria and fed what it found to former Quartz reporter Justin Rohrlich. He used that information as the basis for nine scoops.

Mining the Luanda Leaks with machine learning

Our team at Quartz helped journalists from around the world use machine learning to mine 700,000 documents detailing how Isabel dos Santos siphoned hundreds of millions of dollars out of Angola, We used machine learning to help reporters locate tax documents, board minutes, and other key pages buried in the trove.

Uncovering a precious-metals scheme targeting seniors

I edited this multi-story investigation in which Quartz reporters on the investigations team found that Facebook ads were leading conservative seniors into a precious-metals pitch that cost them thousands of dollars.