Pranav Subramaniam

In today’s organizations, including institutions and companies, more people than ever before rely on insights from data. This increased access to data (data democratization) has increased risk of improper data sharing, data breaches, and inadequate data quality. These challenges have pressured organizations to enforce rules around who can access data, data quality, retention, etc. These rules are often spelled out in a written policy, and enforcing this policy on a storage system requires manual communication between the policy writers (e.g., security/legal experts) and storage system administrators (e.g., DB admins, data engineers, etc.). My PhD work aims to build systems that use NLP techniques, such as LLMs, to accurately audit storage system implementations against written policies to alleviate this type of overhead.