SysteMPC Workshop

The 1st Workshop on Systems for Secure Multi-Party Computation (SysteMPC) was held on July 10, 2025 at the Duan Family Center for Computing and Data Sciences at Boston University. This workshop is organized by Vasia Kalavri, John Liagouris, and Mayank Varia from the BU systems research group and the BU security group.

The SysteMPC workshop brings together cryptographers and systems researchers to discuss advances in overcoming practical challenges of using MPC in the wild. Topics of interest include cryptography and systems co-design, programming abstractions, modularity of cryptographic software, hardware acceleration, experimentation, and integration with the existing ecosystem.

We thank the U.S. National Science Foundation (award #2209194), the BU Computer Science department, and the BU Center for Reliable Information Systems & Cyber Security (RISCS) for their generous support.

Schedule

We have several excellent speakers at the workshop: Marcel Keller and Peter Rindal are giving keynote talks, and we additionally have talks by Marina Blanton, John Liagouris, Peihan Miao, Antigoni Polychroniadou, Mariana Raykova, and Kert Tali along with many lightning talks.

Slides from some of the talks are available in the links below.

Talks

Talk abstracts and bios are listed below.

Marcel Keller – MP-SPDZ: Six Years of Scalable MPC (slides)

Abstract:
While there is a growing number of MPC implementations, most of them are restricted in terms of protocols, security models, applications, and scalability. MP-SPDZ on the other hand offers more than 40 protocol variants in a range of security models and a programming interface that is protocol-independent. I will talk about the core design choices of MP-SPDZ and how they facilitate versatility, user-friendliness, and efficiency. While some of these choices date back to MP-SPDZ’s predecessor, others have been added later. A recent result shows that MP-SPDZ outperforms all considered frameworks that use secret sharing by striking a trade-off between network rounds and RAM usage while preserving usability. This is based on an idea by Büscher et al. (CCS’18). The talk will also highlight more recent additions to MP-SPDZ such as protocols with function-dependent preprocessing and secure shuffling.

Bio:
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.

Kert Tali – Bridging the Gap Between MPC and Information Systems (slides)

Abstract:
While the MPC community has made strides in protocol performance and security, domain-specific languages and compilers, the bottleneck of widespread adoption of MPC has shifted towards systems-level challenges. There is a lack of a common interface and vocabulary when it comes to augmenting information systems with MPC. The MPC technologies on the market today come in a variety of flavors, each introducing a distinct set of concepts and dependencies. We propose the “MPC Engine” abstraction to unify the way how MPC is incorporated within systems and software. This talk will take a look at this paradigm from the viewpoints of analysts and programmers. We will see the common challenges in designing MPC-supported applications based on recent projects, and how these have prompted a redesign of Sharemind MPC to be more modern, integration friendly, and interchangeable with other technologies.

Bio:
Kert is a software architect in the Sharemind MPC product development team at Cybernetica. He was introduced to MPC in 2021 when writing his Master’s thesis on scaling parallel algorithms on MPC. Lately, his focus has been on porting the core of Sharemind MPC to the cloud native Carbyne Stack platform, and designing the JOCONDE MPC-as-a-Service system for official statistics in a joint project between Eurostat and Cybernetica.

Mariana Raykova – Willow: Secure Aggregation Framework with Single-Shot Clients

Abstract:
In this talk we will discuss the challenges for real deployments of private analytics solutions based on our experience with single server and two server solutions in the context of a Federated Learning and Analytics Platform and the Exposure Notifications Private Analytics. We will propose a new framework for construction of secure aggregation solutions, which aims to achieve properties that will facilitate deployments with various underlying trust architectures and also make transitions between those fairly seamless.

Bio:
Mariana Raykova works in the areas of cryptography and security. She is interested in both theoretical work that develops new cryptographic tools and applied cryptography projects that aim to use and implement cryptographic protocols in systems in order to enhance their security properties. Her research includes work in the areas of secure computation, oblivious data structures, zero knowledge and verifiable computation, obfuscation.

She received her PhD from the Computer Science Department of Columbia University, where she was co-advised by Tal Malkin and Steve Bellovin. Then, she spent a year as a postdoc at the Cryptography Group at IBM Research Watson, was a Research Scientist at the Computer Science Laboratory at SRI International between 2013 and 2015, and was an Assistant Professor at the Department of Computer Science at Yale University between 2016 and 2018. She joined Google as a Research Scientist in 2019.

Marina Blanton – Lessons from the Design and Experimentation with the PICCO Compiler (slides)

Abstract:
PICCO is one of the early compilers for transforming a general-purpose program into a corresponding secure multi-party computation protocol and executing it in a distributed environment. It was designed to provide a balanced approach to performance optimization. Since its introduction in 2013, the compiler was extended with dynamic memory management capabilities that permit building any data structure from private or a mix of private and public data. It was also formally analyzed for correctness and security properties.

In this talk, we describe our experience with using PICCO for a variety of application domains and by people with a varying level of expertise. Its current implementation can be used for research experimentation, as a tool for teaching development of MPC programs, and as a software suite for creating deployable MPC applications. We discuss challenges associated with supporting general-purpose programs and the need for intuitive programming abstractions.

Bio:
Marina Blanton is an Associate Professor in the Department of Computer Science and Engineering at the University at Buffalo (UB). She also serves as the Faculty Director of Women in Science and Engineering (WiSE) program at UB. She received her MS degrees from Ohio University and Purdue University and her PhD from Purdue University. Her research interests are centrally in information security, privacy, and applied cryptography and recent projects focus on topics related to secure computation and outsourcing. Dr. Blanton has over 80 refereed publications, has served on the technical program committees of top conferences such as USENIX Security, IEEE S&P, and CCS, and is currently an associate editor of ACM Transactions on Privacy and Security. She has received multiple awards for her research, including a 2013 AFOSR Young Investigator Award, the 2015 ACM CCS Test of Time Award, and a 2018 Google Faculty Research Award.

John Liagouris – The BU Secure Analytics Stack

Abstract:
“The performance of MPC-based approaches is so low that practical applicability is not in sight.” This is a review excerpt of a paper I co-authored, describing our vision to use multiparty computation (MPC) for secure data analytics in the cloud. In this talk, I will share how – 4+ years later – we have realized this unlikely vision and more. I will first explain the legitimate skepticism of the particular reviewer and why past results indicated that MPC protocols were impractical for complex analytics. I will then argue that careful system design and cross-layer optimizations can not only amortize MPC costs, but also achieve scalability to large inputs and complex workloads, without compromising security. I will present the BU Secure Analytics Stack, our unified software architecture for secure collaborative data analysis. Finally, I will show performance results for secure relational and time series analytics at a scale that a few years ago was only possible with information leakage or the use of trusted compute.

Bio:
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). His research interests lie in distributed systems, cloud computing, security and privacy, and data management. Before 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, Greece. John obtained his PhD from NTUA, Greece. His work is supported by a NSF SaTC Core Medium Award, a Bosch Research Award, and a Red Hat Collaboratory Research Incubation Award.

Peihan Miao – Enhancing Private Set Intersection for Broader Applications (slides)

Abstract:
Private set intersection (PSI) enables two parties, each holding a private set of elements, to compute the intersection of their sets without revealing anything beyond the intersection. As a special case of secure multi-party computation, PSI has found many applications and shown early success in practice. In this talk, we will discuss the limitations of standard PSI protocols in terms of functionality and scalability, as well as new models and techniques developed to address these challenges. In particular, we will focus on designing enhanced PSI for two settings: (1) scenarios that require fuzzy/noisy matching, and (2) large-scale streaming data with dynamic and updatable sets.

Bio:
Peihan Miao is an Assistant Professor in the Department of Computer Science at Brown University. Her research interests lie in cryptography, theory, and security, with a focus on secure multi-party computation. She received her PhD from the University of California, Berkeley in 2019. She is a recipient of the NSF CAREER Award, Meta Privacy Enhancing Technologies Award, Google Research Scholar Award, and Amazon Research Award.

Peter Rindal – The Remarkable Success of Pseudorandom Correlation Generators (slides)

Abstract:
Pseudorandom Correlation Generators (PCGs) have emerged as a powerful and versatile tool in the design of efficient secure computation protocols. This talk traces the historical arc of correlated randomness in cryptography—from the foundational roles of Oblivious Transfer and Beaver triples to more structured primitives such as matrix triples. These correlations have long served as the backbone of many secure multi-party computation (MPC) protocols, enabling efficient evaluation of arithmetic and Boolean circuits.

We then shift focus to the development of PCGs: lightweight cryptographic primitives that allow a pair of parties to expand a short seed into a large volume of correlated randomness, without interaction and with strong security guarantees. I will survey key constructions of PCGs for OT correlations and Beaver triples, discuss their surprising efficiency, and highlight how they outperform traditional approaches by several orders of magnitude in both computation and communication.

Finally, we reflect on how the success of PCGs reshapes the cryptographic landscape—suggesting new protocol paradigms, improving precomputation pipelines, and paving the way for highly scalable secure computation.

Bio:
Peter Rindal is a cryptographic researcher specializing in secure multi-party computation (MPC) and privacy-preserving cryptographic protocols. His work focuses on the design of highly efficient protocols for secure computation, with a particular emphasis on practical applications and real-world performance. Peter has contributed to foundational advances in the development of pseudorandom correlation generators (PCGs), as well as optimized constructions for oblivious transfer, Beaver triples, and other forms of correlated randomness. His research bridges the gap between theory and implementation, pushing the boundaries of what is achievable in modern cryptographic systems.

Selected abstracts and bios for lightning talks are listed below.

Christopher Smith – Boots on the Ground with MP-SPDZ: A Developer’s Perspective

Abstract:
In many ways, the MP-SPDZ engine has come very close to turning general purpose MPC into an off-the-shelf tool accessible to the common developer. Prototyping is fast, scripting in Python feels natural, and abstraction comes easy. But some awkward usability quirks remain. For example, there is no branching or indexing on secret values, and there is a lack of support for field embeddings. After a quick peek at the underlying issues, we see that some of these quirks are not native to MP-SPDZ; rather, they seem to arise from more fundamental aspects of the underlying MPC protocols and computational model. This talk offers a field report from hands-on experiences, highlighting what works, what doesn’t, and where there’s room for practically oriented MPC researchers to meet developers where they are.

Bio:
Christopher Smith is a third year PhD candidate in the department of computer science at Stony Brook University, co-advised by Erez Zadok and Omkant Pandey. His current research focuses on applications of MPC and zero-knowledge proofs to the design of secure long-term archival systems, with an emphasis on information-theoretic techniques.

Muhammad Faisal – Building a Secure Machine Learning engine

Abstract:
Secure machine learning using MPC enables a range of applications that require collaborative model training or privacy-preserving inference. For example, companies may collaborate to train ML models that benefit all parties (coopetition), or leverage LLMs to enhance employees productivity. In this talk, we present the design of our secure ML system and its API, highlighting both system-level considerations and user perspective.

Bio:
Muhammad Faisal is a PhD candidate at Boston University, currently in his sixth year. His research focuses on applied cryptography systems, with an emphasis on MPC. His work includes Secrecy (USENIX NSDI 2023) and TVA (USENIX Security 2023).

Tarakaram Gollamudi – ILA: Correctness via Type Checking for Fully Homomorphic Encryption

Abstract:
RLWE-based Fully Homomorphic Encryption (FHE) schemes add some small noise to the message during encryption. The noise accumulates with each homomorphic operation. When the noise exceeds a critical value, the FHE circuit produces an incorrect output. This makes developing FHE applications quite subtle, as one must closely track the noise to ensure correctness. However, existing libraries and compilers offer limited support to statically track the noise. Additionally, FHE circuits are also plagued by wraparound errors that are common in finite modulus arithmetic. These two limitations of existing compilers and libraries make FHE applications too difficult to develop with confidence.
In this work, we present a correctness-oriented IR, Intermediate Language for Arithmetic circuits (ILA), for type-checking circuits intended for homomorphic evaluation. Our IR is backed by a type system that tracks low-level quantitative bounds (e.g., ciphertext noise) without using the secret key. Using our type system, we identify and prove a strong functional correctness criterion for ILA circuits. Additionally, we have designed ILA to be maximally general: our core type system does not directly assume a particular FHE scheme, but instead axiomitizes a model of FHE. We instantiate this model with the exact FHE schemes (BGV, BFV and TFHE), and obtain functional correctness for free. We implement a concrete type checker ILA, parameterized by the noise estimators for three popular FHE libraries (OpenFHE, SEAL and TFHE-rs). We also use the type checker to infer the optimal placement of modulus switching, a common noise management operation. Evaluation shows that ILA type checker is sound (always detects noise overflows), practical (noise estimates are tight) and efficient.

Bio:
Tarakaram Gollamudi works at the intersection of Programming languages and cryptography. He graduated in 2021 with a PhD in Mathematics (Number Theory) from Brandeis. This is a joint work with Anitha Gollamudi and Joshua Gancher.