May 18, 2016;
11:20 a.m.-11:35 a.m.
Brendan Farmer currently works at Panamplify, an early-stage startup that uses AI (automated planning and machine learning) to make insights from large data sets accessible to human analysts. He is also involved in cryptography research in public key infrastructure design, using auditable data structures to solve core problems with digital verification and authentication. Previously, Brendan was a cofounder at NationalField.com, a software startup developed to take insights in organizational management from the Obama for America campaign (2008) and apply them to a variety of contexts, from large healthcare providers (Kaiser Permanente/British NHS) to non-profits (Sierra Club). NationalField grew to millions of users and was acquired in 2013. After his involvement with NationalField, Brendan attended Duke University as an AB Duke Scholar, where he majored in mathematics and philosophy and minored in political science. He's worked as a consultant for large tech companies and startups, including Google, Opower, and Harvard's i-Lab. His long-term goals include further exploring AI in an academic research context, and making the Internet a safer and more trustworthy place through cryptography.
As universities begin to translate more of their student credentials into digital form, being able to efficiently and accurately verify this data in a manner that is agnostic to file format (PDF, XML, etc.) takes on greater importance. We propose a novel method for credential verification based on concepts present in the blockchain, and examine how our solution, centered around cryptographic data structures called transparency logs, can be an easier, more flexible, and more cost-efficient way for universities to digitally verify credentials than existing options such as digitally-signed PDF's. Further, we argue that using transparency logs is a better solution than relying on the blockchain itself, because transparency logs offer many of the benefits of a blockchain-based system without the added complexity. Finally, we look ahead to future possibilities for a system that uses cryptographic data structures for widespread credential verification.