Nikolay Avramov

I am a first-year Ph.D. student at the Privacy Preserving Systems Lab, supervised by Prof. Anwar Hithnawi. I’m broadly interested in data provenance and accountability in machine learning systems. Specifically, my research explores how zero-knowledge proofs can make provenance and auditing practical. At the moment, I focus on techniques for auditing private training datasets, including proving whether a model was trained on a person’s data without revealing the dataset. Before joining the PPS Lab as a Ph.D. student, I was a research assistant there working on zero-knowledge proofs for ML inference. My work included improving the efficiency of ZK inference proofs for a fixed (committed) model, as well as exploring practical proof techniques for collaborative learning settings. I hold an MSc in Computer Science from ETH Zurich and a BSc in Computer Science from the University of Manchester.

Publications

Artemis: Efficient Commit-and-Prove SNARKs for zkML Paper Github

Hidde Lycklama*, Alexander Viand*, Nikolay Avramov, Nicolas Küchler, Anwar Hithnawi

Preprint, arXiv:2409.12055