Description
Did you ever notice that the most important and impactful data sources are inherently sensitive in nature - from healthcare to finance. Even seemingly banal information like your commuting patterns, purchasing habits or insurance reimbursements, can tell a nuanced story of who you are and how you can be influenced. This causes a headache for the honest data scientist who is interested in big picture analytics, not spy-stats.
Fortunately, there has been a huge movement over the past 10-15 years to create a new type of tech stack founded on the concept of privacy-enhancing technologies (PETs), which act as brokers of trust between data sources and data scientists.
In this talk we will give an overview of how we have been building (and today launching) a new python framework, which integrates directly with Jupyter via a backend extension, to make using PETs trivial in the day to day functions of a data scientist. We call it Antigranular.
We expect this project to be a long term endeavour and if you are interested in privacy, confidentiality, security and, of course, data science we would encourage you to get involved.