Presented by:

Billy

Billy VanCannon

Baffle, Inc

Billy VanCannon has over 15 years of experience in cybersecurity. This includes network and software security, public-key infrastructure, and identity and access management. He is currently the Director of Product Management at Baffle, where we work to make encryption easy. Billy has an BS in electrical engineering, an MBA at Northwestern's Kellogg school of management and is CISSP certified.

Bafflelogo 1

Min-Hank Ho

Baffle, Inc.

Min-Hank Ho leads the Product Management team at Baffle. Previously, he led development of the Advanced Security Option for the Oracle database and was one of the original engineer for Transparent Data Encryption.

No video of the event yet, sorry!

Pgvector extends PostgreSQL capabilities to be a vector database as well as an enterprise-class relational database. But, with vector embeddings identified as a potential source of information leakage, the challenge is ensure that companies can take advantage of Postgres and pgvector while remaining compliant with increasingly stringent data security and privacy requirements, especially for GenAI applications.

In this session, we review a first-ever data security solution for pgvector that ensures both the sensitive data and the vector embeddings generated from the data remain protected and inaccessible to any human users (including the Postgres DBA) while all pgvector semantic search operations function correctly. By using RBAC, the protected values will only be in clear text for applications and users that have privileges to access them.

This ability to protect and control access to both the data and its embeddings enables the secure usage of Postgres as a combined data and vector store that simplifies the infrastructure used for GenAI applications and reduces their operational and management.

Date:
Duration:
50 min
Room:
Conference:
Postgres Conference 2025
Language:
Track:
Variants and Cloud
Difficulty:
Medium