Teodora Baluta

I am on the academic job market for positions starting 2024!

Invited Talks: CISPA, Max Planck Institute for Security and Privacy, UC Santa Barbara, Northeastern, Purdue, Georgia Tech, University of Waterloo, Columbia, University of Sydney, Arizona State University.

[CV][Research Statement][Teaching Statement]

I am one of the EECS Rising Stars 2023.

I’m a Ph.D. candidate at National University of Singapore (NUS), advised by Prateek Saxena and Kuldeep S. Meel. I am a part of the KISP lab, as well as MeelGroup.

I work in computer security. My thesis is on Rigorous Security Analyses for Machine Learning (ML) Systems, consisting of the following foundational aspects:

• Definitions: Stochastic gradient descent steps are collision resistant, under precise definitions and mild testable conditions [CCS 23]. This is useful in solving intellectual property disputes.

• Abstractions: We propose a causal model for stochastic gradient descent to analyze the connection between generalization and tests for memorization [CCS 22a].

• Sound procedures for statistical verifiability: black-box [ICSE 21] and white-box [CCS 19].

Broader Interests. I enjoy problems that are both algorithmic and practically relevant. I have also worked on differential privacy for graphs [CCS 21][CCS 22b], learnability of security analysis rules [NDSS 19], improving SAT solving using causality [SAT 23], and cross-language code translation [FSE 21][OOPSLA 23].

My research has been recognized and supported by the Google PhD Fellowship (2021-2023), EECS Rising Stars 2023, Dean’s Graduate Research Excellence Award (2022-2023, Sem.2), and the President’s Graduate Fellowship (2017-2021). I was selected as a finalist for the Microsoft Research PhD Fellowship, Asia-Pacific (2021).

I was a Teaching Assistant @ NUS for several courses: Systems Security CS5231 Fall 2019, Systems Security CS5231 Fall 2018, and Intro to Security CS3235.