Vitaly Shmatikov
More info
- Affiliation
- Cornell Tech
- Known for
- The Netflix de-anonymization, attacks on machine learning
Security researcher with a habit of finding the leak in systems everyone else had already declared safe.
At the University of Texas at Austin he co-wrote the 2008 Netflix paper with Arvind Narayanan. In 2012 came "The Most Dangerous Code in the World", which showed that a long list of libraries and apps used TLS without ever properly checking the certificates — encrypted, yes, but to anyone at all.
At Cornell Tech he turned to machine learning: he co-authored the 2017 membership inference paper, which showed you can tell whether a given record was in a model's training data, and later work on backdoors in federated learning. The theme never changes: data that was supposed to stay in gets out.