CarbyneStackCon '24¶
CarbyneStackCon is the annual gathering of the Carbyne Stack Open Source Community, designed to foster collaboration, discussion, and knowledge sharing on the latest advancements in secure multiparty computation (MPC).
Registration¶
CarbyneStackCon '24 (CSC24) is an open event sponsored by Bosch Research taking place on November 27th (talks) and 28th (workshops), 2024 in Renningen, Germany. Everyone interested in enterprise-grade open MPC is welcome to attend! However, seats for in-person participation are limited and registration is required for both in-person and virtual attendance in order to facilitate our planning.
Important
Please use the event code CSC24 to register on the EventXP platform. On the desktop version of the EventXP website, you must enter it in the the upper right corner of the UI. For the iOS mobile version, you will find a button at the very top. For Android use the hamburger menu button ().
Registration closes on November 20th, 2024.
Program¶
Conference Day (Nov 27)¶
Time | Speaker | Title |
---|---|---|
8:30 am - 9:30 am | - | Registration & Welcome Coffee |
9:30 am - 9:45 am | Dr. Sven Trieflinger | Welcome & Opening Remarks |
9:45 am - 10:15 am | Dr. Marcel Keller | MP-SPDZ After 6 Years |
10:15 am - 10:45 am | Abhilash Venkatesh | TEE-based Secure Computation and its Application to Carbyne Stack |
10:45 am - 11:00 am | - | Coffee Break |
11:00 am - 11:30 am | Dr. Ajith Suresh | Revitalizing Privacy-Preserving Machine Learning: Introducing FANNG-MPC |
11:30 am - 12:00 pm | Kert Tali | Self-Service Deployment in MPC-as-a-Service |
12:00 pm - 1:00 pm | - | Lunch Break & Group Photo |
1:00 pm - 1:30 pm | Dr. Jonas Böhler | Trees & Trade-offs for Secure Predictions |
1:30 pm - 2:00 pm | Dr. Christoph Bösch Prof. Dr. Thomas Hoeren Merlin Rombach | MPC as a Tool for Data Anonymization in Data Analytics |
2:00 pm - 2:30 pm | Adrián Vaca Humanes Daniele Romanini | Building a Privacy-First Ecosystem: SMPC in AdTech Analytics |
2:30 pm - 3:00 pm | Dr. Ian Zhou Vincent Rieder | Silentium: Low Communication & Hardware Acceleration for Beaver Triple Generation |
3:00 pm - 3:15 pm | - | Coffee Break |
3:15 pm - 3:45 pm | Dr. Sven Trieflinger Sebastian Becker Dr. Benjamin Hettwer | Thymus: Adding essential Security Features to the Carbyne Stack platform the Cloud-Native Way |
3:45 pm - 4:15 pm | Dr. John Liagouris | Integrating the BU Secure Analytics Stack with Carbyne Stack |
4:15 pm - 4:45 pm | Dr. Brian LaMacchia Dr. Sven Trieflinger Dr. Christian Hoeppler | Open Source MPC - Quo Vadis |
4:45 pm - 5:00 pm | - | Closing Remarks |
Talk Details¶
9:45 am - 10:15 am
MP-SPDZ After 6 Years
Dr. Marcel Keller (Senior Research Scientist, CSIRO Data61)
Dr. Marcel Keller is a senior research scientist with CSIRO's Data61, a research unit of Australia's national science agency. After completing his PhD with Ivan Damgård at Aarhus University, he spent a few years at the University of Bristol under the supervision of Nigel Smart. There, he started working on an implementation of multi-party computation that eventually would form the basis of MP-SPDZ, an open-source project used by researchers all over the world.
Abstract
MP-SPDZ is the leading open-source research prototype for multi-party computation, a major privacy-enhancing technology. In this talk, I will present more recent developments in MP-SPDZ. On the protocol side, this includes secure shuffling, which enables more efficient data analytics, and the switch to SoftSpokenOT, a more secure and flexible basis for scaling two-party computation. In addition, I will cover MP-SPDZ's approach to scaling computation efficiently, which has been inspired by HyCC (CCS'18).
10:15 am - 10:45 am
TEE-based Secure Computation and its Application to Carbyne Stack
Abhilash Venkatesh (Lead Engineer, CDPG, IISc)
Abhilash is a Lead Engineer at the Centre of Data for Public Good (CDPG), Foundation of Society (FSID), Innovation and Development, IISc. He works primarily on container, cloud-based technologies and Secure Enclave, SMPC-based privacy-enhancing technology (PET) at CDPG's research division. He loves Linux, computer networks, security, privacy, and anything related to systems in general.
Abstract
Trusted Execution Environment (TEE) is a hardware-based technology that provides data integrity, confidentiality, and code integrity during data processing. Commonly used Multi-Party Computation (MPC) paradigms provide privacy-preserving computation via two phases: an input-independent offline phase, which uses heavyweight cryptographic tools to generate cryptographic randomness in advance, and a second online phase consisting of lightweight operations that consume this randomness. This talk will cover the integration of TEE with MPC to securely accelerate the MPC offline phase in collaboration with the Carbyne Stack team at Bosch Research. The talk will also present the applications of TEEs to various privacy-enhancing use cases.
10:45 am - 11:00 am
Coffee Break
11:00 am - 11:30 am
Revitalizing Privacy-Preserving Machine Learning: Introducing FANNG-MPC for Actively Secure MLaaS
Dr. Ajith Suresh (Senior MPC Researcher, Technology Innovation Institute, Abu Dhabi)
Dr. Ajith Suresh is a Senior MPC Researcher at the Technology Innovation Institute (TII) in Abu Dhabi, affiliated with the Cryptography Research Center. Prior to joining TII, he completed 1.5 years of post-doctoral research at the Cryptography and Privacy Engineering (ENCRYPTO) group at the Technical University of Darmstadt, under the supervision of Prof. Thomas Schneider. He holds a PhD from the Indian Institute of Science Bangalore. His research focuses on the design and development of applied Multi-Party Computation (MPC) protocols, with additional interests in privacy-preserving machine learning and federated learning.
Abstract
In response to the reproducibility crisis in scientific research, we present FANNG-MPC, a versatile secure multi-party computation (MPC) framework designed for privacy-preserving machine learning as a service. FANNG is a data-oriented fork of the now-deprecated SCALE-MAMBA, featuring new libraries and instructions optimized for private neural networks. Key innovations include decoupling offline and online phases, a software-based dealer model, and integrated database support. It also introduces advanced protocols for garbled circuits, convolution operations, and private comparisons. This talk will explore FANNG’s design challenges, solutions, and unresolved issues, with a focus on engaging CarbyneStackCon’s community to refine and extend FANNG’s capabilities for privacy-preserving ML.
11:30 am - 12:00 pm
Self-Service Deployment of Computation Tasks in a Multi-Tenant MPC-as-a-Service
Kert Tali (Architect, Sharemind MPC Product Development, Cybernetica)
Kert joined the Sharemind MPC product development team at Cybernetica in 2021 as a programmer. Fascinated by the engineering challenges of MPC deployments, he began exploring ways to enhance the practicality of production-grade MPC. In 2022, Kert defended his Master's thesis on scaling parallel algorithms on MPC. Shortly after, he set his sights on Carbyne Stack. Realizing its immense potential for standardizing the way MPC is deployed, he spearheaded the initiative to integrate Sharemind MPC with the Carbyne Stack platform. As the team grew around this shared ambition, he assumed the role of an architect.
Abstract
Employing MPC for cross-organizational data processing has well-established merits. Yet, the selection of technological solutions that prioritize viability as much as security is limited. The UNECE Input Privacy Preservation Project assessed that servitisation (i.e., MPC-as-a-Service) is key for the widespread adoption of MPC; however, it's not without critical design criteria to compete with data sharing as the preferred method. Such a system would need to propose comprehensive processes and lightweight workflows— accessible to non-technical users for securely and collaboratively organizing and executing MPC tasks. This talk expands on the ongoing work in JOCONDE, a project between Eurostat and Cybernetica to specify an MPCaaS system for use in official statistics. We will explore how general-purpose MPC platforms like Carbyne Stack play a vital role in powering this system.
12:00 pm - 1:00 pm
Lunch Break & Group Photo
1:00 pm - 1:30 pm
Trees & Trade-offs for Secure Predictions
Dr. Jonas Böhler (Lead AI Security & Privacy Researcher, SAP)
Dr. Jonas Böhler is the lead AI security & privacy researcher for the SAP Foundation Model, SAP's table-native AI solution for prediction tasks on tabular data. Previously, he was a senior researcher at SAP Security Research and also serves as SAP's project lead for the EU project Glaciation, focusing on privacy-preserving collaborative learning. Jonas received his PhD from the Karlsruhe Institute of Technology (KIT), where his thesis received awards from the ERCIM Security and Trust Management Working Group and the KIT faculty of computer science. His research interests focus on privacy-enhancing technologies with applications in cross-company collaborations.
Abstract
Privacy-preserving edge-to-cloud data operations are central to the EU project Glaciation. SAP explores collaborative learning across companies, empowered by Carbyne Stack’s cloud-native secure multi-party computation stack. For structured data, as commonly found in industry settings, tree-based models offer simple, efficient, and often accurate prediction solutions. However, secure multi-party training of tree-based models still incurs significant overhead. Fortunately, there exists a spectrum between cloud-outsourced secure multi-party training and edge-local processing. This talk will delve into Glaciation’s progress, discuss secure tree training, and examine various trade-offs considered to optimize the process.
1:30 pm - 2:00 pm
MPC as a Tool for Data Anonymization in Data Analytics: A Legal and Technical Perspective
Dr. Christoph Bösch (Research Engineer, Bosch Research), Prof. Dr. Thomas Hoeren (Head of ITM, University of Münster), and Merlin Rombach (Academic Associate, ITM, University of Münster)
Dr. Christoph Bösch is a research engineer at Bosch Research in the field of security, privacy, and cryptography in general, with a particular focus on applied cryptography and the challenges associated with privacy engineering.
Prof. Dr. Thomas Hoeren is Head of the Institute for Information, Telecommunications and Media Law (ITM) at the University of Münster and current lecturer at the University of Vienna, he is a experienced legal scholar and former judge at the Court of Appeal of Düsseldorf (Copyright Senate). Throughout his career, he has held various prestigious positions and has made significant contributions to the fields of information technology law and intellectual property rights.
Merlin Rombach is a Academic Associate at the Institute for Information, Telecommunications and Media Law (ITM) at the University of Münster. His research interests span cybersecurity law, data protection law, and AI regulation, with his doctoral research specifically exploring how cryptographic methods can enhance personal data protection and potentially even achieve anonymization. Through his work, he examines the legal framework for privacy-enhancing technologies in the digital age.
Abstract
As organizations increasingly rely on data to drive decision-making, handling sensitive information responsibly and in compliance with data protection regulations like the GDPR is essential. Multi-Party Computation (MPC) provides a secure way to process data for analytical purposes while meeting privacy standards. However, companies have been cautious in adopting MPC for real-world applications due to legal uncertainties and unclear benefits. This talk presents the findings from a legal assessment of MPC’s potential as a tool for data anonymization in line with GDPR, with a special focus on HR analytics, where sensitive employee data is involved.
2:00 pm - 2:30 pm
Building a Privacy-First Ecosystem: How SMPC is Transforming AdTech Analytics
Adrián Vaca Humanes (Engineering Lead, Resolve) and Daniele Romanini (Senior Privacy Engineer, Resolve)
Adrián Vaca Humanes is the Engineering Lead at Resolve, focusing on Data Analytics and Cloud Architecture. His experience spans various industries, including banking, consumer discretionary, and telecommunications. At Resolve, he is dedicated to building privacy-preserving solutions for the AdTech industry, with a current focus on data engineering, platform optimization, and cost reduction for products based on Secure Multi-Party Computation (SMPC).
Daniele Romanini is a Senior Privacy Engineer at Resolve, bringing expertise in both data science and software engineering. His background includes experience in academia, government organizations, and the AdTech industry. Daniele is an advocate for privacy-by-design and a privacy tech enthusiast, actively integrating privacy threat modeling and a privacy-first approach into the software development lifecycle. He is currently focused on contributing to the development of a decentralized measurement and analytics platform built with privacy-enhancing technologies at its core.
Abstract
In a rapidly evolving landscape marked by growing user awareness of privacy and increasingly tightening privacy regulations, the AdTech industry faces challenges to remain relevant and profitable. With increased attention to privacy-enhancing technologies, including decentralized computation, Resolve is building decentralized programmatic advertising solutions that emphasize privacy. Secure Multi-Party Computation (MPC) is a fundamental tool in Resolve's arsenal. Specifically, Resolve leverages Carbyne Stack to construct a general-purpose collaborative analytics platform for the AdTech industry. This talk will demonstrate how Carbyne Stack can be applied in programmatic advertising, detailing concrete use cases for collaborative analytics. Additionally, the presentation will dive into technical aspects, addressing scalability and security challenges encountered in real-world deployments.
2:30 pm - 3:00 pm
Silentium: Beaver Triple Generation with Low Communication and Hardware Acceleration
Dr. Ian Zhou (Research Engineer, UTS) and Vincent Rieder (PhD Student, Bosch Research)
Dr. Ian Zhou received his B.S. degree in computer science from The University of Sydney in 2016 and completed his M.B.A. and Ph.D. degrees at the University of Technology Sydney in 2019 and 2023, respectively. His Ph.D. research focused on machine learning-based frost monitoring systems. Currently, Dr. Zhou is a Research Engineer specializing in privacy-preserving technologies, working on accelerating multi-party secure machine learning algorithms with CUDA and other GPU platforms. His research interests include AI, IoT, cyber-physical systems, and blockchain.
Vincent Rieder is a PhD student at Bosch Research in Renningen, focusing on Secure Multi-Party Computation (MPC). Holding a Master’s degree in mathematics, his research centers on the algorithmic aspects of MPC. He aims to introduce new protocols into the Carbyne Stack MPC cloud platform, specifically by optimizing and implementing an enhanced MPC offline phase for generating Beaver triples using a Pseudorandom Correlation Generator (PCG) with low communication overhead, named Silentium.
Abstract
The MPC framework of Carbyne Stack relies on the SPDZ protocol, where the most resource-intensive task is the generation of Beaver triples in the offline phase. A recent advancement for offline phases, particularly relevant for the cloud context, involves Pseudorandom Correlation Generators (PCGs) with reduced communication demands. This talk introduces Silentium, a PCG implementation designed to generate Beaver triples with hardware acceleration. The first part of the presentation will cover Silentium’s design principles, highlighting how this approach outperforms previous offline phases in MP-SPDZ. The second part will focus on how further enhance Silentium’s local phase with a GPU-optimized Number Theoretic Transform for large degrees. Altogether, this talk discusses how Silentium can enhance Klyshko, the offline phase engine of Carbyne Stack, potentially offering significant cost savings in future applications.
3:00 pm - 3:15 pm
Coffee Break
3:15 pm - 3:45 pm
Thymus: Adding essential Security Features to the Carbyne Stack platform the Cloud-Native Way
Dr. Sven Trieflinger (Senior Project Manager, Bosch Research), Sebastian Becker (Research Engineer, Bosch Research), and Dr. Benjamin Hettwer (Research Engineer, Bosch Research)
Sebastian Becker is a Research Engineer at Robert Bosch GmbH. His work focuses on making privacy-enhancing technologies easily adaptable to the needs of the wide range of application domains at Bosch. In this context, Sebastian also works as one of the maintainers of and main contributors to Carbyne Stack.
Dr. Benjamin Hettwer is a research engineer at Robert Bosch Corporate Research in Renningen, Germany. He received his Ph.D. in Electrical Engineering and Information Technology from Ruhr-Universität Bochum, Germany, in 2020. His research interests focus on the intersection of hardware security and machine learning. Recently, he also started working on privacy-enhancing technologies such as the secure multi-party computation solution Carbyne Stack and its application to industrial use cases.
Dr. Sven Trieflinger is a Senior Project Manager, Group Manager for Security, Privacy, and Safety, Research Engineer, and open source software maintainer at Bosch Research. He has over 15 years of experience in the design, architecture, and implementation of distributed systems and cloud platforms. With his team at Bosch, Sven drives innovation in the area of privacy-preserving computing technologies and is spearheading open source computing on encrypted data technology with the Carbyne Stack cloud-native Secure Multiparty Computation platform.
Abstract
There is a new kid on the block: Thymus adds eagerly awaited security-related capabilities to the Carbyne Stack platform. In this talk, we will share, how we are providing versatile authentication via Ory Kratos, flexible authorisation based on the Open Policy Agent, and security for inter-VCP communication channels and user-facing APIs via Istio Kubernetes-native network machinery. We will discuss the rationale behind technology choices and provide insights into how to use the new features.
3:45 pm - 4:15 pm
Integrating the BU Secure Analytics Stack with Carbyne Stack
Dr. John Liagouris (Assistant Professor, Boston University)
Dr. John Liagouris is an assistant professor of Computer Science at Boston University, where he co-leads the Complex Analytics and Scalable Processing research lab (CASP). He is a member of the Systems Group and is also affiliated with the Security Group at BU. John's research focuses on distributed systems, cloud computing, security & privacy, and data management. Prior to joining BU, he was a visiting scholar at the RISELab, UC Berkeley, a senior researcher at the Systems Group, ETH Zurich, a visiting research fellow at the University of Hong Kong (HKU), and a research assistant at the “Athena” Research Center in Greece. John earned his PhD from NTUA, Greece. His work has received several awards, including an “Outstanding New Research Direction Award” at Usenix HotStorage 2020, a NSF SaTC Core Medium Award, a Bosch Research Award, and a Red Hat Collaboratory Research Incubation Award.
Abstract
In this talk, Dr. John Liagouris will present the latest developments in the BU secure analytics stack and its integration with Carbyne Stack. The BU stack is a novel software stack developed at Boston University to facilitate general-purpose analytics using secure Multiparty Computation (MPC). Its current version supports both relational and time series computations on millions of input records, with configurable semi-honest or malicious security settings. Built from scratch, the BU stack encompasses efficient implementations of low-level MPC functionalities up through high-level operators and programming abstractions. All stack layers, except the bottom one, are protocol-agnostic, employing a hierarchical and modular design that maximizes reusability and extensibility. Improvements in any component of the stack can cascade across other parts, enhancing overall functionality. This talk will delve into the BU stack's 4-year research journey, detailing the innovation led by the Systems Group and Security Group at BU.
4:15 pm - 4:45 pm
Open Source MPC - Quo Vadis
Dr. Brian LaMacchia (Executive Director, MPC Alliance), Dr. Sven Trieflinger (Senior Project Manager, Bosch Research), and Dr. Christian Höppler (Open Source Officer, Bosch Research)
Dr. Brian LaMacchia is an applied cryptographer and currently the Executive Director of the MPC Alliance, a consortium of over 50 organizations promoting secure multi-party computation (MPC) technology. After a 25-year career at Microsoft Corporation, where he served as Distinguished Engineer for Cryptography and led the Security and Cryptography team at Microsoft Research, Brian retired in December 2022. He also co-founded and chaired the Microsoft Cryptography Review Board. Beyond his role at the MPC Alliance, Brian is an Adjunct Associate Professor at Indiana University-Bloomington’s School of Informatics and Computing, an Affiliate Faculty member at the University of Washington’s Department of Computer Science and Engineering, and an Advisor to Quantropi, Inc. Brian currently serves as Treasurer for the International Association for Cryptologic Research (IACR) and as Vice President on the Board of Directors of Seattle Opera. He earned his S.B., S.M., and Ph.D. degrees in Electrical Engineering and Computer Science from MIT in 1990, 1991, and 1996, respectively.
Dr. Sven Trieflinger is a Senior Project Manager, Group Manager for Security, Privacy, and Safety, Research Engineer, and open source software maintainer at Bosch Research. He has over 15 years of experience in the design, architecture, and implementation of distributed systems and cloud platforms. With his team at Bosch, Sven drives innovation in the area of privacy-preserving computing technologies and is spearheading open source computing on encrypted data technology with the Carbyne Stack cloud-native Secure Multiparty Computation platform.
Dr. Christian Höppler is Open Source Officer at Bosch Research and member of Bosch's corporate Open Source Expert Team. Since joining Bosch Research a decade ago Chris has been working on a wide range of open source topics from compliance to strategy. Currently, he focuses on working with internal teams to help with their open source contributions both within Bosch Research and beyond. To support that work and to reduce some of the friction encountered he has begun to automate the contribution lifecycle. Additionally, Chris has been a member of the research project "Economy of Things", which is working towards well-governed decentralized systems for a digital economy. Prior to that he's been working in the automotive sector, mainly developing test automation software for hardware-in-the-loop testbenches.
Abstract
Three years ago, we set sail on a quest to create the most advanced cloud-native Secure Multiparty Computation (MPC) platform. Our goal was to plant a seed for an open ecosystem where state-of-the-art MPC technology can thrive. Building Carbyne Stack “in the open” has been a deliberate choice and open source software plays a pivotal role as it fosters collaboration and transparency, allowing for continuous peer review and rapid identification of vulnerabilities. As we approach a state where Carbyne Stack implements all the essential features for deployment in real-world use cases, we believe it's time to rethink how we work together and decide on what a happy home for the Carbyne Stack community should look like. In this tag-team talk, Brian will discuss the history and importance of open source in the development of cryptographic software, then Sven and Chris will relate what Brian discusses to the history of Carbyne S tack and shed some light on the way forward for the initiative.
4:45 pm - 5:00 pm
Closing Remarks
Workshop Day (Nov 28)¶
9:00 am - 10:30 am
Donating Carbyne Stack to an Open Source Foundation
Details
Carbyne Stack is coming of age. Initially launched as a Bosch Research initiative, we think it is time to transition the project to a community-driven model. This workshop will explore the benefits, potential foundation options, and the process required for such a donation. Key discussion points include:
- Advantages of transitioning Carbyne Stack to an open source foundation
- Selection of the most suitable foundation
- Steps and tasks involved in the donation process
- Roles and responsibilities of funders post-donation
- Governance structure for a community-led project
10:30 am - 11:00 am
Coffee Break
11:00 am - 12:30 pm
Generalizing Carbyne Stack for Multi-Runtime / Multi-Protocol MPC Support
Details
Carbyne Stack is currently tightly integrated with MP-SPDZ as the underlying MPC engine, specifically in a malicious majority setting. This approach prioritizes security, but it may not always be feasible due to performance and cost considerations. To maximize the platform's value, this workshop will discuss ongoing efforts by various partners to support additional protocols and explore how to create a unified, more flexible framework that supports multiple MPC engines and protocols.
12:30 pm - 1:30 pm
Lunch Break
1:30 pm - 3:00 pm
Walk-in Topics
Details
Open time for additional discussions, spontaneous topics, and networking.
Venue Information¶
-
Bosch Research Campus
Address
Robert-Bosch-Campus 1
71272 Renningen, Germany
Getting There¶
The Bosch Research Campus is located near Stuttgart. See our travel information sheet for information on how to get to the Research Campus by car or public transport.
Accommodation¶
There are numerous hotels in and around Renningen. One within walking distance to the venue is:
Hotel Campo
Raitestraße 26
71272 Renningen
+49 7159 939800
info@campo-renningen.de
Catering¶
Food and beverages during the event will be complementary.