Verifiable and post-quantum distributed learning
Letting federated learning prove its results with zero-knowledge proofs, keeping it secure against quantum attacks, and verifying models trained across institutions.
Research
We make distributed and federated AI more secure, private, and checkable. Right now we focus on zero-knowledge and post-quantum methods for federated, decentralized, and edge learning.
Resilience against malicious and strategic participants
Defenses that discourage data poisoning and gaming the system, plus decoys that make attacks more expensive to run.
Bayesian incentive mechanism · Blockchain honeypot conversion
Cryptographic assurance for emerging systems
Hardware fingerprints and zero-knowledge proofs that let a device show it is patched and trustworthy, with cryptography light enough for small devices.
Methods. Applied cryptography, secure and distributed systems, adversarial machine learning, incentive-aware design, and reproducible systems measurement, selected according to the threat model.
Applications. IoT and edge, cyber-physical and critical infrastructure, healthcare, and digital-asset ecosystems.