Sponsored by: Capital One | Secure Data Collaboration: A Modern Enterprise Enclave Architecture
Overview
| Experience | In Person |
|---|---|
| Track | Data Sharing & Collaboration |
| Industry | Enterprise Technology, Healthcare & Life Sciences, Financial Services |
| Technologies | Delta Sharing, Unity Catalog, Databricks Apps |
| Skill Level | Intermediate |
| DOWNLOAD SESSION SLIDES | |
Collaborating on sensitive data is critical for business insights but creates significant security and compliance risks. This session introduces a secure data enclave architecture that enables cross-functional analysis on sensitive enterprise data (like PII or IP). Learn how this model strengthens data privacy and provides a comprehensive, auditable trail without moving raw data.
Session Speakers
Vince Goveas
/Director, Product Management
Capital One
Full Summary
Secure data collaboration without exposing raw data
Enterprises in regulated sectors hold valuable but siloed datasets. The highest value insights often require joining data across organizations, yet privacy mandates, competitive sensitivities, and governance rules make traditional sharing risky. A new approach, introduced jointly by Capital One Software and Databricks and available through the Databricks Marketplace, enables partners to collaborate on tokenized data inside a controlled clean room so raw values never leave the source. Organizations navigating these challenges can find a practical starting point in security best practices for the Databricks Platform, which outlines the governance controls that underpin safe, scalable data sharing.
FAQ
A clean room provides shared infrastructure, access controls, and collaboration tools for multiple parties. It still requires loading sensitive data into a common environment. Tokenizing sensitive fields before data enters the clean room closes that gap by reducing exposure while preserving the ability to analyze and link records.
Each party replaces sensitive elements with tokens before sharing. The tokens are designed to be linkable across datasets, so collaborators can find matches and run analytics without reversing tokens to their original values.
No. Data preparation uses your native Databricks compute, and linking happens on compute within the clean room infrastructure in your Databricks account. Collaboration remains within your organizational boundary.
It was introduced jointly with Databricks and is available through the Databricks Marketplace. Teams can explore it there and provide feedback.
Examples include building a 360-degree patient view to improve care and research, reducing months of manual data preparation for analysts and scientists, and accelerating innovation and revenue by enabling partnerships that were previously stalled by privacy and compliance concerns.