AA Andreea B. Alexandru

Andreea B. Alexandru

Researcher

Selected publications

CRYPTO 2025General Functional Bootstrapping using CKKS

Brings functional bootstrapping to CKKS, so arbitrary functions can be evaluated during the refresh on encrypted real-valued data.

PNAS 2023Collaborative Privacy-Preserving Analysis of Oncological Data using Multiparty Homomorphic Encryption

Encrypted analysis of real oncological data across institutions, a landmark deployment of multiparty homomorphic encryption.

SODA 2025Sublinear-Round Broadcast without Trusted Setup

Byzantine broadcast in a sublinear number of rounds with no trusted setup, against a dishonest majority.

ASIACRYPT 2022State Machine Replication under Changing Network Conditions

State-machine replication that stays secure and fault-optimal even as the network shifts between synchronous and asynchronous.

IEEE CSM 2021Encrypted Control for Networked Systems: An Illustrative Introduction and Current Challenges

An accessible introduction to running control loops on encrypted data and the open challenges of the field.

IEEE TAC 2020Cloud-based Quadratic Optimization with Partially Homomorphic Encryption

A foundational framework in control for solving quadratic optimization problems on encrypted data in the cloud.

Research overview

Functional bootstrapping for CKKS

My current research centers on fully homomorphic encryption in its many flavors, including its threshold variants. I am particularly interested in developing and optimizing functional bootstrapping for the CKKS scheme.

Bootstrapping is the operation that refreshes a ciphertext so that an unbounded number of homomorphic operations can be performed on it; without it, computation on encrypted data is limited to a fixed, shallow depth. Functional bootstrapping goes one step further: instead of merely refreshing the ciphertext, it evaluates an arbitrary function during the refresh, turning an expensive but unavoidable maintenance step into useful computation.

Bringing this capability to CKKS, the scheme best suited to encrypted real-valued and machine-learning workloads, makes non-polynomial operations such as comparisons, activation functions, and table lookups practical at scale, which is a key enabler for efficient privacy-preserving analytics and machine learning on encrypted data.

From this line of work Functional bootstrapping for CKKS · CRYPTO 2025 Larger lookup tables · TCHES 2026 Sparse Hermite interpolation · ASIACRYPT 2026
CKKS functional bootstrapping: refreshing a ciphertext while evaluating a function

Bringing FHE to real-world deployments

A large part of my current research aims to bring privacy-preserving technologies to real-world deployment. Using and extending open-source libraries such as OpenFHE and NVFlare, I build privacy-preserving pipelines for tasks such as secure data querying and evaluation, including private information retrieval; scalable and secure collaboration for encrypted analytics over federated medical data; and efficient conversions (transciphering) between homomorphic encryption and symmetric ciphers.

Alongside performance, I study the security of approximate homomorphic encryption and how to configure it correctly for real deployments. This work is carried out at Duality Technologies.

From this line of work Encrypted oncological analytics · PNAS 2023 Application-aware FHE · CiC 2026 OpenFHE · ePrint 2022
Privacy-preserving collaborative analytics: institutions hold encrypted local data with threshold key shares while a server computes on ciphertexts

Secure multi-party computation and distributed protocols

During my postdoc, my research interests were in secure multi-party computation and distributed cryptographic protocols such as broadcast and consensus. Although this line of work has already spanned decades, it has gained tremendous interest with the emergence of blockchains and, more generally, with the scaling up of distributed systems.

My collaborators and I designed efficient protocols for state machine replication and broadcast under various threat models and network-synchronicity assumptions. For instance, arbitrary changes in network synchronicity can render a protocol designed for synchronous networks insecure, while a protocol designed for asynchronous networks tolerates fewer faults even when the network is synchronous. I proposed common-subset and state-machine-replication protocols that have low communication complexity and tolerate the optimal number of faults even under arbitrary network transitions from synchronous to asynchronous.

I was also interested in robust differential privacy and anonymous communication systems, as well as in combining fully homomorphic encryption with complementary tools such as garbled circuits for private anomaly detection.

From this line of work State machine replication · ASIACRYPT 2022 Sublinear-round broadcast · SODA 2025 Private anomaly detection · CDC 2022
Consensus and state-machine replication with Byzantine parties under changing network conditions

Privacy and security of dynamical systems

During my PhD, my main focus was the privacy and security of dynamical systems. Many of the structures around us are dynamical systems, from the time series of medical monitoring sensors to the energy consumption of our homes, and many abstract processes can be modeled as such, including iterative optimization algorithms like gradient descent and the training and evaluation of neural networks.

I designed privacy-preserving control and optimization algorithms built on homomorphic encryption and secure multi-party computation, which compute directly on encrypted data so that private information is not leaked to the computing party. Concealing dynamical data brings challenges beyond the static case: dependencies between data at different time steps, maintaining privacy across consecutive iterations, and accumulation of noise in the result.

Privacy-preserving control and optimization across cloud, vehicles, hospitals, homes and buildings

My PhD projects spanned:

From this line of work Encrypted quadratic optimization · IEEE TAC 2020 Encrypted LQG · ICCPS 2019 Motion planning with secrecy · ACC 2019 Private data-driven control · CDC 2020 Encrypted control, an introduction · IEEE CSM 2021

Looking ahead

I plan to pursue several research directions:

All publications

Full list on Google Scholar

Conferences

Journals and book chapters

Technical reports and whitepapers

Preprints

Invited talks, workshops and posters