Mingshen Sun
TikTok's Privacy Innovation
#1about 2 minutes
Understanding the challenges of secure data collaboration
Data collaboration platforms must balance data provider control with consumer access while protecting the confidentiality and integrity of data.
#2about 1 minute
Establishing goals for a modern data collaboration framework
A successful framework must prioritize usability, security, privacy enforcement, data accuracy, and ease of cloud deployment.
#3about 3 minutes
Limitations of existing data privacy solutions
Existing solutions like SQL-based clean rooms, differential privacy, and TEEs each fail to meet all requirements for usability, accuracy, and privacy.
#4about 3 minutes
Introducing a two-stage data clean room solution
A novel two-stage approach separates interactive programming on privacy-enhanced data from execution on full data within a secure environment.
#5about 2 minutes
How trusted execution environments enable secure collaboration
TEEs provide hardware-based isolation and attestation reports, guaranteeing code and data integrity for all parties in a collaboration.
#6about 2 minutes
Applying the platform to research and machine learning
The platform serves as a trusted research environment (TRE) and supports use cases in advertising analytics and privacy-preserving machine learning.
#7about 4 minutes
Demonstrating the platform's workflow and attestation
A live demo shows how a researcher uses a Jupyter Notebook to train a model and verifies its execution in a TEE via a downloadable attestation report.
#8about 2 minutes
Learnings and future roadmap for the platform
Key takeaways include using TEEs for privacy, and the future roadmap involves multi-way collaboration, multi-cloud support, and GPU integration.
#9about 1 minute
Announcing the open source release on GitHub
The data collaboration platform is now open source, allowing developers to experiment with it and contribute to the project.
Related jobs
Jobs that call for the skills explored in this talk.
Featured Partners
Related Videos
23:52Safeguarding Sensitive Data Access At Scale with Privacy-Enhancing Technologies
Mingshen Sun
59:34This Machine Ends Data Breaches
Liz Moy
30:23The Impact of Technology on Democracy
Fanny Jimenez, David Simas, Constantijn van Oranje & Vera Futorjanski
29:31Break the Chain: Decentralized solutions for today’s Web2.0 privacy problems
Adam Larter
23:50Data Privacy in LLMs: Challenges and Best Practices
Aditi Godbole
24:33GenAI Security: Navigating the Unseen Iceberg
Maish Saidel-Keesing
29:00Beyond the Hype: Building Trustworthy and Reliable LLM Applications with Guardrails
Alex Soto
27:34Confuse, Obfuscate, Disrupt: Using Adversarial Techniques for Better AI and True Anonymity
David vonThenen
From learning to earning
Jobs that call for the skills explored in this talk.
TikTok Shop - Data Scientist - Governance and Experience Data Science
TikTok
Charing Cross, United Kingdom
Python
Data analysis
Machine Learning
Security-by-Design for Trustworthy Machine Learning Pipelines
Association Bernard Gregory
Machine Learning
Continuous Delivery
Senior Data Analyst, Governance & Experience - TikTok Shop EMEA
TikTok
Charing Cross, United Kingdom
Senior
Python
Data analysis
Audit Manager - Data, Privacy & AI
Edinburgh, United Kingdom
Senior Data Analyst - TikTok LIVE - London
TikTok
Charing Cross, United Kingdom
Senior
Python
Tableau
A/B testing
Data analysis
VDI Endpoint Security Specialist - USDS
TikTok
Charing Cross, United Kingdom
Intermediate
iOS
VMware
Android
vSphere
Powershell
+2