Duality Tech Talks

Use AWS Nitro Enclaves and Duality’s Privacy Protected AI Solution to futureproof, secure, and scale data acquisition for model development, tuning, testing, and customization while protecting model IP and maintaining the privacy of data inputs.

The recent executive order from the US Whitehouse and the current NIST RFI for guidelines in developing and using AI securely while maintaining the privacy of data inputs have raised many questions. Fortunately, the combination of AWS’s Nitro Secure Enclaves with Duality’s Privacy Protected AI Solution provides critical answers to tough data questions facing AI Engineering Teams: How do we acquire data at the scale and diversity needed to move models quickly to production? How can we collaborate with 3rd parties to validate and tune our model on real data versus mock or synthetic data? How do we coordinate with data from multiple parties? How do we prove to clients that our model is valuable to them? How do we do all that while protecting our IP and the privacy of data?

Join Duality’s Head Data Scientist and AI Engineer, Zohar Duchin, and Sr Partner Solution Architect, Ashok Mahajan of AWS, to hear how this joint solution can help AI Engineering teams overcome data acquisition, training, inference, customization, and monetization challenges while protecting model IP and the privacy of data inputs.

Duality's Zero Footprint Investigation and Intelligence solutions utilizes fully homomorphic encryption to satisfy cross-border and cross-domain data sharing security, privacy, and legal obligations by default. This solution greatly benefits those needing answers to questions from 3rd party, sensitive data sets: KYC, AML, investigations, intelligence activities. Learn how it works from system requirements to setup and use.

This session includes:
- Overview of common use cases
- Infrastructure requirements
- Data set support (structured and unstructured)
- Demo explaining and showing how it works
- Q&A

Governments around the world share a common goal: make better use of data by increasing data sharing while increasing data protections.

The traditional means of satisfying regulations and agency policies are typically done with checkpoints to anonymize or otherwise obfuscate both the data and the use of the data. These traditional methods strain budgets, slow progress, and limit the accuracy of data insights.

Instead, Duality offers quantum-secure data guardrails that accelerate and grow the speed and scale of sensitive data operations while also eliminating as much as 90% of costs for high-security environments.

This means faster and greater collaboration across agencies, with coalition partners, and streamlined use of PAI/CAI while maintaining Operatonal Security.

"How can we customize models on sensitive client data that we can't have while protecting our model IP?"

Today, we'll show you how our Confidential AI solution operationalizes advanced technologies and techniques (confidential computing, federated learning) into a plug-and-play software solution that AI and data teams can trust, and use.

Many organizations today have been developing proprietary models (e.g., risk models, predictive models) that are useful in guiding internal decisions and improving client services. Today, these clients are demanding customized models using their sensitive enterprise data.

Unfortunately, traditional methods of protecting data and software aren't enough to turn this demand into a new, scalable stream of revenue.

Join us as we discuss and show how our clients are accelerating time to market for ML/AI focused initiatives.

Protect the data. Protect the model. Leap ahead of the competition.

Learn how Mastercard has taken the lead in effective anti Fraud, KYC, and AML efforts with a privacy protected workflow via Duality (as published by the IMDA).

Fighting financial crimes is a data game. Crime fighters need answers quickly and must ensure their investigations remain confidential, as not to leak their progress to suspected organizations or individuals. However, criminals continue to evade authorities and slip through the cracks despite efforts around KYC and AML.

Why? Because data protection requirements slow and limit what data can be shared, and with whom. The criminals know this and hide behind our own processes, and they're winning.

However, those hindrances are eliminated when advanced encryption for data in-use is operationalized into these processes.

The result is a near-instant, self-service means of gathering and using information instead of waiting for weeks or months for an answer about a transaction, identity, or other.

See how Mastercard and other financial institutions have upleveled their internal KYC and AML efforts while also improving their response time when collaborating on government investigations.

Workloads shown: secure query (all data types)
Workloads discussed: query, analytics, machine learning
Technologies used: fully homomorphic encryption (FHE), confidential computing, federated learning
Solution shown: Zero Footprint Investigations
Solutions discussed: Confidential AI & Zero Footprint Investigations