WorldFestival 2022 WorldFestival 2022

Data Protection with ML on Device

- PDT
mobileWeek Main Stage
Join on Hopin

Giorgio Natili
Capital One, Director of Engineering

On paper, I am a Director of engineering at Capital One where I lead the team the integrates insightful experiences and personalizations in all the supported surfaces for all the Capital One customers. In the past, I had the privilege to lead the Kindle rendering team solving the problem of delivering millions of books on billions of devices in every supported language. Also, I worked as an engineering lead in Amazon Payments and implemented payments methods and regulations for Brazil.

In a parallel universe, I foster my vision of an inclusive and diverse community where great training is affordable to everyone. To achieve this goal, I run several conferences (Droidcon Boston and SwiftFest), participate in many meetups in Boston, Seattle, and New York, and speak at conferences covering my favorite topics: Android, Angular, Machine Learning, Swift and TDD.

I have the honor of being part of the Google Developer Experts program. 

Sudheendra K. Kaanugovi
Capital One, Engineering Leader and Mobile Architect

Master of Technology in Embedded Systems, Bachelor of Technology in Electronics and Communication Engineering
Shipping award winning enterprise level Mobile Applications from past 13 years. Extensive knowledge of architecting, developing and publishing applications that are Full Native, Hybrid and Web Technologies. Currently venturing into Machine Learning Technologies with targeted interest focusing on small factor devices. In my free time I’d like to work on building custom Home Automation devices and writing software. 


When collecting analytics to feed data pipelines that enable monitoring and alerts for your digital property, you are exposed to the risk of transmitting and storing sensitive data. Even more, if a user inputs data into a form, it's hard to detect if they submitted sensitive data.

Detecting and masking sensitive data before they leave a digital property minimizes the risk of data leaking and protects the identity of your customers.

During this talk, we will discuss how to implement a solution that leverages machine learning to protect customers' Personally Identifiable Information (PII) on Android/iOS (~97% detection accuracy). After an architecture overview and a discussion on the alternatives evaluated, we will dig into the platform-specific implementations to then discuss the drawbacks and opportunities to expand the solution.