About Me
Hello everyone, welcome to my website!
My name is Donya, I am a Ph.D. candidate at TTIC (it's a wonderful place to focus on Machine Learning theory at UChicago). My research lies at the intersection of machine learning and theoretical computer science, with a focus on well-defined aspects of safety and trustworthiness in machine learning, including robustness, reliability, and explainability with provable guarantees. I develop algorithmic and statistical frameworks to study these properties.
Humbled by their capabilities, I’ve become increasingly interested in the agentic view of models. I began my Ph.D. research under the supervision of Avrim Blum in March 2024.
I had the pleasure of spending half of the summer 2025 at Carnegie Mellon University, hosted by Andrej Risteski, and the other half at Columbia University, hosted by Daniel Hsu. I will be a long-term student participant in Simons Institute's Federated and Collaborative Learning program in Winter 2026 - shoot me an email if you'll be around 🙂. I'm very grateful for all my incredible mentors and the opportunities.
During my undergrad, I interned at Max Planck Institute, Germany, under the supervision of Peter Benner, and the University of Texas at Austin, USA. I had the opportunity to get to know fantastic people from all around the world and learn from them. I was fortunate to work on my Bachelor's Thesis under the supervision of Keshav Pingali from the University of Texas at Austin and was the first (and hopefully not the last) Bachelor's student in my university who has officially done her thesis internationally.
You can find my CV here .
Publications
Most of my publications have authors ordered alphabetically, indicated by (α–β), as is customary in theoretical computer science.
|
Taming Imperfect Process Verifiers: A Sampling Perspective on Backtracking Dhruv Rohatgi, Abhishek Shetty, Donya Saless, Yuchen Li, Ankur Moitra, Andrej Risteski, Dylan J. Foster Preprint |
|
Prior Makes It Possible: From Sublinear Graph Algorithms to LLM Test-Time Methods Avrim Blum, Daniel Hsu, Cyrus Rashtchian, Donya Saless(α–β) Preprint |
|
Proofs as Explanations: Short Certificates for Reliable Predictions Avrim Blum, Steve Hanneke, Chirag Pabbaraju, Donya Saless(α–β) COLT 2025 |
|
PAC Learning with Improvements Idan Attias, Avrim Blum, Keziah Naggita, Donya Saless, Dravyansh Sharma, and Matthew Walter(α–β) ICML 2025 |
|
Regularized Robustly Reliable Learners and Instance Targeted Attacks Avrim Blum, Donya Saless(α–β) ALT 2026 NeurIPS 2025 Workshop: Reliable ML from Unreliable Data Workshop |
|
HyperNetVec: Fast and scalable hierarchical embedding for hypergraphs Sepideh Maleki, Donya Saless, Dennis P. Wall, Keshav Pingali NetSci-X 2022 ICML 2021 Workshop: Self-Supervised Learning for Reasoning and Perception |