Michelle Brown

About Michelle Brown

Practitioner in the field of democracy assistance with nearly two decades of experience in strengthening electoral integrity. • Election expert with an extensive network at the intersection of elections and technology. • Polymath with the curiosity and aptitude for learning new technologies, their core functions, and using them to address democratic issues. • Data scientist with experience applying AI, statistical, and computer vision methods to address real-world problems and challenges. • Communicator with ability to convey complex technical concepts and ‘translate’ between disparate communities such as democracy practitioners, diplomats, political scientists, programmers, political leaders, and non-governmental organizations. • Strategic advisor to 16 different civil society organizations during high pressure situations during more than 40 elections in places such as Egypt, Indonesia, Kenya, Nigeria, Myanmar, Ukraine, and Serbia. Strategic Thinking & Thought Leadership Advances NDI’s mission and the global dialogue within the democracy promotion space through detecting and leveraging opportunities for innovation. Track record of identifying cross-cutting trends and topics in electoral integrity, such as advocating for open election data, increasing information integrity, and protecting against state surveillance. Thinks about how interventions can address and incorporate local voices and needs. Advises the Global Network of Domestic Election Monitors (GNDEM) on short- and longer-term strategies. Tech Overseen the development of products such as: Factalyzer, which allows organizations to analyze social media and uncover networks of inauthentic accounts; Foresight 5.0, which allows organizations to use sampling and reporting to evaluate an election and project results; Basic Biometric Box, which uses sensors to illustrate the software & hardware issues of biometric ID; Godo, a messaging chatbot helping observers send photos of the results forms; and Scroller, which was a Chrome extension that collected ad information on Facebook. Machine Learning, Deep Learning, Artificial Intelligence Led pilots showing how: an election statement is generated by retraining a GPT-2 model; NLP techniques can assess a larger amount of social media data than humans alone; computer vision can surface videos most likely to contain fraud; and image similarity algorithms can identify networks of accounts. Co-author of “Hidden in Plain Sight?” article, which uses a CNN to quickly analyze forms to identify irregularities (Electoral Studies 74: 102411)

Experience

1. Senior Advisor on Elections & Director of the Open Election Data Initiative

   National Democratic Institute (NDI)

   Jan 2008 — Present · Washington, DC, US

Education

• Georgetown University

  Master of Public Policy (MPP), emphasis on Quantitative Methods

• University of Arizona

  Bachelor's degree, Political Science & Digital Media (Double Major)