Cloud computing systems have quickly become ubiquitous as a way to deliver a variety of applications to users in a convenient and flexible package. However, these distributed systems also pose new security challenges. Unexpected complications or gaps can appear between different parts of a distributed system, allowing an attacker to perform actions that each part of the system allows but that achieve an undesirable result as a whole. The goal of our project is to build a tool that observes normal use patterns of a system, detects malicious variations from those patterns, and intervenes appropriately

In 2009, I tested a distributed system used for a column-oriented database while interning at Thomson Reuters. More recently, I’ve built a prototype app for a human-computer interaction research project under Professor Charles Oman in the MIT Humans and Automation Lab, and I’ve worked on backend data structures at Google.