Pasadena, CA / email@example.com / http://abelson.live
Experienced Data Scientist and Full-Stack Engineer, Open Source Software Developer, Digital Journalist, and Civic Hacker with a track record of using data for social good.
Kickstarter (Senior Data Engineer, 2017-present)
Rearchitected Kickstarter’s aging data infrastructure in two months, moving to a combination of Airflow, DBT, Redshift Spectrum, and Fivetran, increasing cluster performance. availability, and analytic insight.
In less than a month, singlehandedly designed, prototyped, and deployed a spam classifier capable of greater than 98% accuracy and close to a 0% false positive rate, ameliorating one of the worst user experiences on the site.
Developed a timeseries algorithm for predicting the likelihood that a Kickstarter project will succeed or fail and how much it will raise (>90% accuracy one day after launch). By constructing the model as an HTTP service, its output has been incorporated in domains as diverse as revenue forecasting, search ranking, customer support, and social media marketing.
Actively participated in company-wide efforts to enforce GDPR guidlines by auditing multiple terabytes of data for PII. Wrote custom software to anonymize sensitive information and implemented standards for data retention.
Rearchitected Kickstarter's user-facing tools for reporting fraud and other violations using React and GraphQL, effectively increasing the granularity and accuracy of user-generated data while retaining the overall volume of reports.
Provided daily support, guidance, and mentorship to three Data Analysts.
Vox Media (Chief Data Scientist, 2016)
Responsible for the design, development, and ongoing maintenance of a centralized Hive warehouse for all of the company’s key data sources, including Google Analytics Premium (BigQuery), Chartbeat, DoubleClick for Publishers (DFP), Chorus (Vox's proprietary content management system), Facebook Insights, Twitter Analytics, Youtube Analytics, Snapchat Analytics, and several others.
Wrote Hive queries and custom Map-Reduce jobs to join and analyze these data sources. These analyses were presented directly to the CEO and were fundamental in steering company-wide revenue strategy and identifying opportunities for future growth.
Enigma (Chief Data Scientist, 2013-2015)
Designed and developed a global database of government procurement contracts within ten non-english speaking countries for a Fortune 100 company. This involved extracting structured data from messy websites, PDFs, Word Documents, PowerPoints, and poorly-formatted Excel spreadsheets. To accomplish this task, I hired and trained a team of four people, designed custom, reusable software to expedite the data acquisition process, and interfaced directly with the clients. The successful delivery of this application was fundamental in securing a multi-year, multi-million dollar contract which increased Enigma’s revenue ten-fold and aided in raising $27 million in venture capital.
Worked with the City of New Orleans to develop machine learning-based tool to assist in addressing the city’s backlog of blight abatement cases. Within three months of the tool’s implementation, the backlog of cases had been fully eliminated. This project recieved an award from the Bureau of Governmental Research.
In collaboration with the Red Cross, DataKind, and local fire departments, I developed a novel machine learning model capable of predicting the likelihood that residents in a given census block lacked a smoke alarm. This project was powered by multiple open-source libraries and resulted in a free web application which enabled local governments to upload their own fire incident data and return a list of addresses to target for inspections. The tool is actively used by multiple fire departments throughout the country, as well as by the Red Cross to optimize their Home Fire Preparedness Outreach Campaign.
Analyzed and visualized 50 years of daily temperature readings from NOAA to identify daily temperature anomalies. The resulting map and analysis helped communicate how climate change results in not only in warmer weather, but more anomalous weather. This visualization generated significant traffic to Enigma’s corporate website, received multiple awards, was presented to the UN’s Intergovernmental Panel on Climate Change, was tweeted by Bill Gates, and was also deemed the “best visualization of climate change” by Jer Thorpe (a prominent data artist).
NewsLynx (Co-founder and CTO, 2013-2015)
DataKind (DataCorps Ambassador, 2013-2014)
New York Times / Knight-Mozilla OpenNews (Data Scientist / OpenNews Fellow, 2013)
Treasury.IO (Technical Lead, 2013)
Harmony Institute (Data Scientist, 2012)
Columbia University (Research Assistant, 2011)
Harry Frank Guggenheim Foundation (Assistant Program Officer, 2009-2011)
Grassroots Campaigns (Assistant Office Manager, 2004)
Columbia University, New York, NY
Whitman College, Walla Walla, WA
Keller, Michael and Abelson, Brian (2015). Newslynx: A tool for newsroom impact measurement. Tow Center for Digital Journalism, Columbia University.
Varshney, Kush R., Abelson, Brian, et. al. (2014). Targeting villages for rural development using satellite image analysis. Big Data 3:1.
Dmochowski, J.P., Abelson, Brian, et. al. (2014). Audience preferences are predicted by temporal reliability of neural processing. Nature Communications 5:4567.