Parnian Kassraie

I am a research scientist at Google DeepMind, working on foundational research for post-training. I am generally interested in optimization, sampling, and statistics, often in the context of reinforcement learning and guided generation.

I completed my Ph.D. in Computer Science at ETH Zurich, advised by Andreas Krause and Peter Bühlmann. My research was generously supported by a Google PhD Fellowship, and collected in a thesis titled: Learning to Optimize in Structured Enviroments. While in graduate school, I interned at Apple MLR and Google DeepMind, mentored by Marco Cuturi and Olivier Bachem, respectively. I also spent a semester at Carnegie Mellon University as a research scholar hosted by Aaditya Ramdas. You can find my CV here [updated September 2025].

News

• [Jul-25] Passed my Ph.D. exam with Kevin Jamieson and Aldo Pacchiano kindly serving as the committee.
• [Jun-25] Organizing the Two Faces of Optimization workshop at ETH Zurich.
• [Apr-25] Organizing the EXAIT workshop at ICML 2025.
• [Oct-24] Received a G-Research PhD Prize in Mathematics.
• [Sep-24] Was awarded a Google Ph.D. Fellowship in Machine Intelligence. [ETH media]
• [Aug-24] Finished my internship at Apple MLR (Paris) with a short talk on my project, Progressive Entropic OT.
• [Mar-24] Was awarded the ETH Doc.Mobility Fellowship, for a research visit with Aaditya Ramdas.
• [Nov-23] Instance-Dependent Generalization Bounds via OT to appear in JMLR.
• [Jun-23] Serving as the head teaching assistant for Probabilistic Artificial Intelligence.
• [Jun-21] Started a PhD.