Experience

Meta (Facebook) | Senior Software Engineer - ML (Sep 2021 - Present)

Currently on the Facebook (FB) Marketplace Integrity anti-scam team

• I create and maintain Spam Folder triaging methods that stops hundreds of thousands of inauthentic messages per day, preventing millions of dollars of scam against users each month
• Implemented a novel technique for identifying scammers whose claimed and actual location do not match. I incorporated the signal into multiple levers in Marketplace Integrity, with significant, demonstrated impact on scammer identification, and it is now being used by multiple teams across Meta.
• I was the Engineering lead (supporting 6 engineers plus 10+ partners and stakeholders) for a short-term, intensive sprint where we successfully blocked the vast majority of seller scammers in one country as proof-of-concept of several experimental tools. We are now expanding many of these ideas into new markets or as full global launches.
• In addition to helping guide the direction of several projects on our team, my individual contributions have led to tens of percent of measurable decrease in scam volume across Marketplace

Was the owner of a large feature (100M+ daily impressions) that supported the FB group ecosystem by providing personalized recommendations of lesser-known groups

• The feature increased the number of FB Groups with sustainable levels of engagement by hundreds of thousands (my contributions increased this effect by 5%)
• Increased click rate by 10%

Supported FB’s comment-writing assistant system

• Improved filtering to increase system-wide coverage by 3%, resulting in millions of additional comments made daily
• Integrated this ML infrastructure into FB’s Birthday Center

FINRA | Data Scientist (Oct 2020 - Sep 2021)

• Developed and tested novel Deep Learning market surveillance models using Tensorflow/Keras
• Extracted useful, model-ready datasets using PySpark
• Queried from petabytes of data – approximately 100B records per day
• Focused on deployment, preparing pipeline for highly scaled production environment
• Built beautiful, interactive visualizations to display model output for business users

TPC Energy Fund | Director of Data Science (May 2020 - Oct 2020)

• Provided highly accurate forecasts of returns and volatility of the wholesale power market using machine learning
• Optimized portfolios across thousands of trades using industry best-practices
• Implemented robust backtesting systems to measure performance and risk
• Customized industry standard solutions to fit nuanced financial situations

DataRobot | Customer Facing Data Scientist (Aug 2019 - Apr 2020)

• As one of the first full-time members of the AI for Good team, Paul was instrumental in shaping the program
• Contributed to strategic direction, and vetted dozens of applications for the first cohort of beneficiaries.
• Represented the program internally to new hires, and externally in media interviews.
• Served as primary CFDS for several high-profile, non-profit partners, including Kiva, DonorsChoose, and Anacostia Riverkeeper, as well as for some traditional paying customers.

Deloitte Consulting | Data Scientist/Senior Consultant (Aug 2014 - Aug 2019)

• Paul of Deloitte Consulting’s Strategy and Analytics Government and Public Sector (GPS) practice.

Data Scientist, Team Lead | Large Business Issue Recommender | Large Dept of Treausury Agency

• Led a team (4 full-time Data Scientists plus 2 part-time senior advisors) that used a recommendation engine approach to perform unsupervised anomaly detection for a large Dept of Treasury Agency
• Paul stood up the project, turning a small code base into an end-to-end data pipeline, complete with a backtesting methodology to benchmark their hundreds of experiments.

Data Scientist, Workstream Lead | Workforce Optimization Tool | Large Dept of Treausury Agency

• Led a team of 3-4 full-time consultants in overhauling and improving a tool to guide workforce allocation responsible for several hundreds of millions in generated revenue
• The tool provided a boost in efficiency worth an estimated tens of millions.
• A user-friendly interface gave business users access to a complex model with ML and optimization components.
• Paul’s work included management of the client relationship, coding, regression modeling, linear optimization, and development of a front-end web-based user interface in R Shiny.

Analytics Consultant | Large Dept of Homeland Security Agency

• Part of the core Analytics team supporting a large 40+ person contract that helped the agency prepare for and remediate findings from their annual
• Created a tool that brought together 10+ data sources that had not previously been connected or reconciled.
• Through dozens of highly complex SQL queries, created a system to automatically flag a variety of issues or potential issues, informing a financial statement adjustment in the tens of millions of dollars.

Training and Analytics Leadership

• Over the years, Paul played several rolls internally at Deloitte around training new practitioners.
• Directly taught hundreds of students internally, dozen externally (to paying clients), and his courses have reached even more. He created several courses in R and SQL, and helped facilitate several others
• Advised curriculum development as Deloitte stood up its internal “Analytics University” for GPS practitioners.

Sr Research Assistant | Federal Reserve Board of Governors (Jun 2012 - Jul 2014)

• Implemented the ideas of Board economists, from data collection and cleaning, to running regressions and models, to producing charts and figures used in academic talks and in policy presentations to the governors and other officers.
• Contributed to research direction and helped to determine model validity.
• Topics varied, but focused largely on how the Fed’s policy actions affect the primary and secondary housing finance markets.

Teaching Assistant and “Learning Assistant” | University of Colorado Applied Math Department (Aug 2009 - May 2012)

• Lectured weekly workshops for 40 engineering students.
• Created a curriculum to follow the professor’s primary lecture, wrote quizzes, and graded homework, quizzes and exams.
• Courses included: Calculus I, Linear Algebra, and Applied Probability

Intern | Fast Enterprises (Jun 2011 - Aug 2011)

• Helped implement Fast core software to fit the specific needs of the State of Montana.
• This included testing, debugging, helping state employees use our software, and coding in SQL and Visual Basic.

Research Intern | Nanyang Technological University (Jun 2010 - Aug 2010)

• Analyzed the changing correlation relationships between different Japanese industries.
• Responsibilities included writing code to solve complex mathematics problems, improving the efficiency of existing code, and automating several processes to make the project more user-friendly.
• Resulted in a co-authored publication.