Hey, I’m Avi! I’m currently a 3rd year undergraduate student at the University of California, Berkeley, where I am a double major in Computer Science and Applied Mathematics. I plan to apply for Ph.D programs for the Fall 2025 class. I’m broadly interested in both machine learning applications for computer security, and security/robustness of machine learning models themselves. Right now, I’m particularly focused on adversarial attacks on language models and the effects of data poisoning on model robustness.

Research

My recent projects include SSH anomaly detection using keystroke inter-arrival times, creating universal transfer attacks on black-box language models, and adversarial curriculum learning for bus-bunching mitigation.

I’m extremely fortunate for the opportunity to conduct research with amazing mentors Julien Piet and Chawin Sitawarin at Berkeley as a part of David Wagner’s group. See my research page for more information.

Other

Previously, I’ve worked at Millennium Management as a quantitative research intern, where I developed a ML-based anomaly detection tool for market data and analyzed trading signals. I will be returning there for Summer 2024. I have also worked under Kalyan Veeramachaneni as a member of MIT’s Data to AI Lab, where I built an open source tool to fit ML pipelines for daily anomaly detection on time-series datasets.

In my free time, I enjoy reading (sociology and fantasy!), weightlifting, and playing the piano.